{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "9ade2e6e",
   "metadata": {},
   "source": [
    "# Ch 02 — End-to-End ML Project (notebook)\n",
    "\n",
    "`[← 01 ml-landscape]` · **this notebook** · `[03 classification →]`\n",
    "\n",
    "Runs top-to-bottom in ~2 min on free Colab CPU. Last verified 2026-06-11.\n",
    "\n",
    "**What you'll build**\n",
    "- One project, start to finish: California district statistics in, a dollar prediction out, saved as a single file you reload and call on a new district.\n",
    "- A `Pipeline` + `ColumnTransformer` leakage firewall that imputes, scales, and one-hot encodes without ever letting test data touch a fitted statistic.\n",
    "- A stratified train/test split, three competing model families judged on one harness, a tuned random forest, and a bootstrap confidence interval on the test metric.\n",
    "- A leakage bug you watch fabricate a great-looking score on a pure-noise target, then fix in one structural move.\n",
    "\n",
    "**How this notebook works.** Code cells with a `# TODO` are yours to fill in. Run the cell to grade yourself: `[ ok ]` passed, `[FAIL]` shows what went wrong, `[ -- ]` means not attempted yet. Every exercise has a hint ladder (open only as many rungs as you need) and a folded solution below it. The notebook runs top-to-bottom even if you fill in nothing: the solution cells redefine the functions so the later parts still work. See Ch 00 for the full protocol.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "98aadf3c",
   "metadata": {},
   "source": [
    "## Before you start\n",
    "\n",
    "Three questions. Each is answerable after Ch 00 and Ch 01. Answer them before you run a cell.\n",
    "\n",
    "1. You split a dataset 80/20, fit a `StandardScaler` on the full data first, then split. What did you just leak, and into where? <details><summary>Answer</summary>The scaler learned the mean and std from all rows, including the 20% that becomes your test set. The test set's information leaked into the training-time statistics. Your test score is now optimistic and you can't trust it. The whole \"Prepare the data\" part exists to make this leak impossible by construction.</details>\n",
    "2. A model reports a cross-validation RMSE of 0.49 and a test RMSE of 0.47. Is that a problem? <details><summary>Answer</summary>No. Test is slightly *better* than CV, which is normal sampling noise when the splits are clean. A test score much *better* than CV is the alarm: it usually means leakage. A test score much *worse* than CV means overfitting to the validation folds or a distribution shift between split halves.</details>\n",
    "3. Predict before you run: of California Housing's nine features, which single one will dominate the correlation with house value? <details><summary>Answer</summary>Median income (`MedInc`), at about +0.69. Everything else is below 0.16 in magnitude. Income dominating the signal is the most important thing the EDA tells you, and it is why we stratify the split on income.</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a367846",
   "metadata": {},
   "source": [
    "## Setup\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "8513a635",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:44.672510Z",
     "iopub.status.busy": "2026-06-12T09:33:44.672436Z",
     "iopub.status.idle": "2026-06-12T09:33:45.308422Z",
     "shell.execute_reply": "2026-06-12T09:33:45.307955Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "numpy 2.2.6 · pandas 2.3.3 · sklearn 1.7.2 · scipy 1.15.3\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import sklearn\n",
    "import scipy\n",
    "print(f\"numpy {np.__version__} · pandas {pd.__version__} · sklearn {sklearn.__version__} · scipy {scipy.__version__}\")\n",
    "if np.__version__ < \"2.0\":\n",
    "    print(\"WARN: this notebook is written for NumPy 2.x; older versions may differ slightly\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "25216689",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.309694Z",
     "iopub.status.busy": "2026-06-12T09:33:45.309563Z",
     "iopub.status.idle": "2026-06-12T09:33:45.313707Z",
     "shell.execute_reply": "2026-06-12T09:33:45.313306Z"
    }
   },
   "outputs": [],
   "source": [
    "import os, random\n",
    "SEED = 0\n",
    "FAST = bool(os.environ.get('NB_FAST'))   # CI smoke mode: fewer trees/search iters, same code paths\n",
    "rng = np.random.default_rng(SEED)         # the one RNG we pass around for reproducible synthesis\n",
    "random.seed(SEED)\n",
    "\n",
    "# ── house self-check harness (identical across all chapter notebooks) ──\n",
    "import numpy as _np\n",
    "\n",
    "def check(label, test_fn, required=False):\n",
    "    \"\"\"Run one self-check. test_fn raises AssertionError (with a teaching\n",
    "    message) on failure, NotImplementedError if the stub is unfilled.\n",
    "    required=True is used only in solution cells; it is what CI grades.\"\"\"\n",
    "    try:\n",
    "        test_fn()\n",
    "    except NotImplementedError:\n",
    "        if required:\n",
    "            raise AssertionError(f\"{label}: reference solution incomplete\")\n",
    "        print(f\"[ -- ] {label}: not attempted yet — fill in the TODO above, then re-run.\")\n",
    "        return False\n",
    "    except AssertionError as e:\n",
    "        if required:\n",
    "            raise\n",
    "        print(f\"[FAIL] {label}: {e}\")\n",
    "        return False\n",
    "    print(f\"[ ok ] {label}\")\n",
    "    return True\n",
    "\n",
    "def attempted(*vals):\n",
    "    \"\"\"Treat None placeholders as 'not attempted'.\"\"\"\n",
    "    if any(v is None for v in vals):\n",
    "        raise NotImplementedError\n",
    "\n",
    "def check_shape(x, want):\n",
    "    assert tuple(x.shape) == tuple(want), \\\n",
    "        f\"shape {tuple(x.shape)}, expected {tuple(want)} — check your reshape/transpose order\"\n",
    "\n",
    "def check_close(got, want, atol=1e-5, rtol=1e-4, msg=\"\"):\n",
    "    g, w = _np.asarray(got, dtype=float), _np.asarray(want, dtype=float)\n",
    "    assert g.shape == w.shape, f\"shape {g.shape} vs expected {w.shape}. {msg}\"\n",
    "    bad = ~_np.isclose(g, w, atol=atol, rtol=rtol)\n",
    "    assert not bad.any(), \\\n",
    "        f\"{bad.mean():.2%} of values wrong (max diff {abs(g - w).max():.3g}). {msg}\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4f0d9ba0",
   "metadata": {},
   "source": [
    "> **Note:** seeds make this notebook's printed numbers reproduce on CPU. Library versions, the number of BLAS threads, and the random-forest implementation can shift the last digit or two; the quoted numbers hold for the pinned environment. If your test RMSE is 0.472 and the page says 0.473, you did nothing wrong. We quote RMSE in the dataset's native units, where the target is median house value in units of \\$100,000 (so 0.47 is about \\$47,000).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0cd5daf9",
   "metadata": {},
   "source": [
    "## The map\n",
    "\n",
    "> **Part 1 — Frame and get the data.** Turn \"predict house prices\" into a task with an input, an output, a metric, and a baseline; load California Housing idempotently and look at its shape.\n",
    "> **Part 2 — Explore (EDA).** Find the four weird things about this dataset in fifteen minutes: the caps, the skew, the geography, and which feature carries the signal.\n",
    "> **Part 3 — The split is a decision.** Build a stratified train/test split on binned income and prove it beats a random split on bias; then lock the test set away.\n",
    "> **Part 4 — The pipeline is the leakage firewall.** Compose imputation, scaling, and one-hot encoding into one `Pipeline` + `ColumnTransformer` that cannot leak, and watch a leak fabricate a score when you step outside it.\n",
    "> **Part 5 — Select, tune, and analyze.** Compare three model families on one harness, tune the winner, read the residuals, and look at where it fails.\n",
    "> **Part 6 — Touch the test set once.** Compute the test metric with a bootstrap confidence interval, save the whole pipeline as one artifact, and reload it to predict on a new district.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a889676c",
   "metadata": {},
   "source": [
    "## Part 1 — Frame and get the data\n",
    "\n",
    "> **Objectives**\n",
    "> - State the task as input, output, metric, and baseline before writing model code.\n",
    "> - Load California Housing through an idempotent, offline loader (no mutable URL).\n",
    "> - Confirm the shape and dtypes match the data dictionary.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "284ad074",
   "metadata": {},
   "source": [
    "### 1.1 Framing the problem\n",
    "\n",
    "Three questions, answered before any model. They take thirty minutes and save weeks.\n",
    "\n",
    "1. **Objective.** Predict the median house value of a California district, accurate enough to flag mispriced listings in an automated valuation pipeline.\n",
    "2. **Metric and tolerance.** Median absolute error in the target's native units (\\$100k) — the median of the per-district absolute errors, robust to the capped-price districts a few wildly-wrong predictions cannot drag it upward. The product team tolerates about \\$25k (0.25 in \\$100k units); better is the goal. We also track RMSE, which squares errors and so weights the capped-price tail more heavily; the two metrics disagree, and reading both is the point.\n",
    "3. **Baseline.** A team of human appraisers, entering prices by hand, with a reported **median absolute error of \\$30k** (0.30 in \\$100k units). Beating \\$30k means we contributed something. (We also keep a dumber, secondary floor — a constant predictor that always guesses the training mean — as a sanity check, but it is the naive predictor, not the bar that matters.)\n",
    "\n",
    "This fixes the rest of the notebook: the task is **supervised regression**, the headline metric is **median absolute error** (with RMSE as a tail-sensitive companion), and the bar to clear is the human-appraiser median AE of \\$30k. Because the baseline is stated as a *median AE*, the only model number that legitimately compares against it is the model's own median AE — not its RMSE or MAE. If you cannot answer these three about your own project, stop coding and answer them first.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3efd6996",
   "metadata": {},
   "source": [
    "### 1.2 Get the data, idempotently\n",
    "\n",
    "`fetch_california_housing` ships with scikit-learn and caches under `data/`. The first call may download a few hundred KB once; every call after is a no-op read from disk. This is the Tier-2 dataset pattern: a built-in loader with an immutable source, not a raw file pulled from a mutable git branch.\n",
    "\n",
    "> **Caveat:** the chapter prose fetched `ageron/data@main/housing.tgz`. Fetching from a branch (`@main`) is a moving target: the file can change under you and there is no checksum. We use the sklearn loader instead. It carries the same 20,640 districts and the 1990 census features, minus the `ocean_proximity` column and the missing values, which we reconstruct deterministically in 1.3 so the imputation and one-hot lessons stay real.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b4f86f75",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.314495Z",
     "iopub.status.busy": "2026-06-12T09:33:45.314427Z",
     "iopub.status.idle": "2026-06-12T09:33:45.343545Z",
     "shell.execute_reply": "2026-06-12T09:33:45.342965Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape (20640, 9)\n",
      "['MedInc', 'HouseAge', 'AveRooms', 'AveBedrms', 'Population', 'AveOccup', 'Latitude', 'Longitude', 'MedHouseVal']\n"
     ]
    }
   ],
   "source": [
    "from sklearn.datasets import fetch_california_housing\n",
    "\n",
    "raw = fetch_california_housing(as_frame=True, data_home=\"data\")  # idempotent: cached under data/ after first call\n",
    "housing = raw.frame.copy()\n",
    "print(f\"shape {housing.shape}\")\n",
    "print(housing.columns.tolist())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8ba9833e",
   "metadata": {},
   "source": [
    "The nine columns and the target:\n",
    "\n",
    "| Column | Meaning |\n",
    "|---|---|\n",
    "| `MedInc` | median income in the district (tens of thousands of USD) |\n",
    "| `HouseAge` | median house age in years (capped at 52) |\n",
    "| `AveRooms` | average rooms per household |\n",
    "| `AveBedrms` | average bedrooms per household |\n",
    "| `Population` | district population |\n",
    "| `AveOccup` | average household occupancy |\n",
    "| `Latitude` / `Longitude` | district centroid |\n",
    "| `MedHouseVal` | TARGET: median house value in units of \\$100,000 (capped at 5.0) |\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "30fd3799",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.344561Z",
     "iopub.status.busy": "2026-06-12T09:33:45.344483Z",
     "iopub.status.idle": "2026-06-12T09:33:45.351256Z",
     "shell.execute_reply": "2026-06-12T09:33:45.350753Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>MedInc</th>\n",
       "      <th>HouseAge</th>\n",
       "      <th>AveRooms</th>\n",
       "      <th>AveBedrms</th>\n",
       "      <th>Population</th>\n",
       "      <th>AveOccup</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>MedHouseVal</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8.3252</td>\n",
       "      <td>41.0</td>\n",
       "      <td>6.984127</td>\n",
       "      <td>1.023810</td>\n",
       "      <td>322.0</td>\n",
       "      <td>2.555556</td>\n",
       "      <td>37.88</td>\n",
       "      <td>-122.23</td>\n",
       "      <td>4.526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8.3014</td>\n",
       "      <td>21.0</td>\n",
       "      <td>6.238137</td>\n",
       "      <td>0.971880</td>\n",
       "      <td>2401.0</td>\n",
       "      <td>2.109842</td>\n",
       "      <td>37.86</td>\n",
       "      <td>-122.22</td>\n",
       "      <td>3.585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7.2574</td>\n",
       "      <td>52.0</td>\n",
       "      <td>8.288136</td>\n",
       "      <td>1.073446</td>\n",
       "      <td>496.0</td>\n",
       "      <td>2.802260</td>\n",
       "      <td>37.85</td>\n",
       "      <td>-122.24</td>\n",
       "      <td>3.521</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   MedInc  HouseAge  AveRooms  AveBedrms  Population  AveOccup  Latitude  \\\n",
       "0  8.3252      41.0  6.984127   1.023810       322.0  2.555556     37.88   \n",
       "1  8.3014      21.0  6.238137   0.971880      2401.0  2.109842     37.86   \n",
       "2  7.2574      52.0  8.288136   1.073446       496.0  2.802260     37.85   \n",
       "\n",
       "   Longitude  MedHouseVal  \n",
       "0    -122.23        4.526  \n",
       "1    -122.22        3.585  \n",
       "2    -122.24        3.521  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "housing.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d6f3402d",
   "metadata": {},
   "source": [
    "> **Interpretation.** Nine numeric features and one numeric target, 20,640 rows. No string columns yet and no missing values yet. Both of those are realities of messy data, so the next cell puts them back in on purpose, deterministically, so the pipeline we build has something to impute and to encode.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "37bbb461",
   "metadata": {},
   "source": [
    "### 1.3 Reconstruct the messy parts (deterministically)\n",
    "\n",
    "The original Géron CSV had a categorical `ocean_proximity` and a column with missing values. The sklearn version is clean. We add both back with the seeded RNG so the result is identical on every run and the EDA, imputation, and one-hot steps all have real work to do. We derive `ocean_proximity` from longitude bands (the coast runs down the west edge of the state, so longitude is a defensible distance-to-ocean proxy) and knock out about 1% of `AveBedrms` to NaN.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ade4f58d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.351995Z",
     "iopub.status.busy": "2026-06-12T09:33:45.351920Z",
     "iopub.status.idle": "2026-06-12T09:33:45.356431Z",
     "shell.execute_reply": "2026-06-12T09:33:45.356077Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ocean_proximity\n",
      "INLAND        7811\n",
      "<1H OCEAN     6901\n",
      "FAR INLAND    5080\n",
      "NEAR BAY       848\n",
      "\n",
      "AveBedrms missing: 204 of 20640\n"
     ]
    }
   ],
   "source": [
    "# derive a categorical from geography: longitude bands as a coast-distance proxy\n",
    "lon_edges = [-125.0, -122.5, -120.5, -118.0, -113.0]   # west (coast) -> east (inland)\n",
    "lon_labels = [\"NEAR BAY\", \"<1H OCEAN\", \"INLAND\", \"FAR INLAND\"]\n",
    "housing[\"ocean_proximity\"] = pd.cut(housing[\"Longitude\"], bins=lon_edges, labels=lon_labels)\n",
    "\n",
    "# knock ~1% of AveBedrms to NaN so the imputer has a job (seeded, so it is reproducible)\n",
    "miss_mask = rng.random(len(housing)) < 0.01\n",
    "housing.loc[miss_mask, \"AveBedrms\"] = np.nan\n",
    "\n",
    "print(housing[\"ocean_proximity\"].value_counts().to_string())\n",
    "print(f\"\\nAveBedrms missing: {int(housing['AveBedrms'].isna().sum())} of {len(housing)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79cc71d9",
   "metadata": {},
   "source": [
    "> **Interpretation.** Four geographic categories with a heavy class imbalance (`NEAR BAY` is rare), and about 204 missing `AveBedrms`. The categorical needs one-hot encoding and the NaNs need imputation, and both must be learned from training data only. That is exactly what the pipeline in Part 4 enforces.\n",
    "\n",
    "> **Key takeaways**\n",
    "> - Framing (objective, metric, baseline) comes before code; it decides everything downstream.\n",
    "> - Idempotent loaders with immutable sources, never a mutable branch URL.\n",
    "> - Real datasets have categoricals and gaps; we reconstructed them deterministically so the lessons are not theater.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1a8a180a",
   "metadata": {},
   "source": [
    "## Part 2 — Explore (EDA)\n",
    "\n",
    "> **Objectives**\n",
    "> - Read `info()` and `describe()` for dtypes, gaps, and caps.\n",
    "> - See the skew in the histograms and the geography in the scatter.\n",
    "> - Rank features by correlation with the target and confirm income dominates.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fbd89fd6",
   "metadata": {},
   "source": [
    "### 2.1 Structure: dtypes, gaps, ranges\n",
    "\n",
    "`info()` tells you dtypes and where the gaps are. `describe()` tells you the ranges, and the ranges are where the surprises hide.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "332578f1",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.357264Z",
     "iopub.status.busy": "2026-06-12T09:33:45.357192Z",
     "iopub.status.idle": "2026-06-12T09:33:45.361271Z",
     "shell.execute_reply": "2026-06-12T09:33:45.360829Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 20640 entries, 0 to 20639\n",
      "Data columns (total 10 columns):\n",
      " #   Column           Non-Null Count  Dtype   \n",
      "---  ------           --------------  -----   \n",
      " 0   MedInc           20640 non-null  float64 \n",
      " 1   HouseAge         20640 non-null  float64 \n",
      " 2   AveRooms         20640 non-null  float64 \n",
      " 3   AveBedrms        20436 non-null  float64 \n",
      " 4   Population       20640 non-null  float64 \n",
      " 5   AveOccup         20640 non-null  float64 \n",
      " 6   Latitude         20640 non-null  float64 \n",
      " 7   Longitude        20640 non-null  float64 \n",
      " 8   MedHouseVal      20640 non-null  float64 \n",
      " 9   ocean_proximity  20640 non-null  category\n",
      "dtypes: category(1), float64(9)\n",
      "memory usage: 1.4 MB\n"
     ]
    }
   ],
   "source": [
    "housing.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9babed12",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.362241Z",
     "iopub.status.busy": "2026-06-12T09:33:45.362166Z",
     "iopub.status.idle": "2026-06-12T09:33:45.376085Z",
     "shell.execute_reply": "2026-06-12T09:33:45.375663Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>MedInc</th>\n",
       "      <th>HouseAge</th>\n",
       "      <th>AveRooms</th>\n",
       "      <th>AveBedrms</th>\n",
       "      <th>Population</th>\n",
       "      <th>AveOccup</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>MedHouseVal</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>20640.00</td>\n",
       "      <td>20640.00</td>\n",
       "      <td>20640.00</td>\n",
       "      <td>20436.00</td>\n",
       "      <td>20640.00</td>\n",
       "      <td>20640.00</td>\n",
       "      <td>20640.00</td>\n",
       "      <td>20640.00</td>\n",
       "      <td>20640.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>3.87</td>\n",
       "      <td>28.64</td>\n",
       "      <td>5.43</td>\n",
       "      <td>1.10</td>\n",
       "      <td>1425.48</td>\n",
       "      <td>3.07</td>\n",
       "      <td>35.63</td>\n",
       "      <td>-119.57</td>\n",
       "      <td>2.07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.90</td>\n",
       "      <td>12.59</td>\n",
       "      <td>2.47</td>\n",
       "      <td>0.48</td>\n",
       "      <td>1132.46</td>\n",
       "      <td>10.39</td>\n",
       "      <td>2.14</td>\n",
       "      <td>2.00</td>\n",
       "      <td>1.15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.50</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.33</td>\n",
       "      <td>3.00</td>\n",
       "      <td>0.69</td>\n",
       "      <td>32.54</td>\n",
       "      <td>-124.35</td>\n",
       "      <td>0.15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>2.56</td>\n",
       "      <td>18.00</td>\n",
       "      <td>4.44</td>\n",
       "      <td>1.01</td>\n",
       "      <td>787.00</td>\n",
       "      <td>2.43</td>\n",
       "      <td>33.93</td>\n",
       "      <td>-121.80</td>\n",
       "      <td>1.20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>3.53</td>\n",
       "      <td>29.00</td>\n",
       "      <td>5.23</td>\n",
       "      <td>1.05</td>\n",
       "      <td>1166.00</td>\n",
       "      <td>2.82</td>\n",
       "      <td>34.26</td>\n",
       "      <td>-118.49</td>\n",
       "      <td>1.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>4.74</td>\n",
       "      <td>37.00</td>\n",
       "      <td>6.05</td>\n",
       "      <td>1.10</td>\n",
       "      <td>1725.00</td>\n",
       "      <td>3.28</td>\n",
       "      <td>37.71</td>\n",
       "      <td>-118.01</td>\n",
       "      <td>2.65</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>15.00</td>\n",
       "      <td>52.00</td>\n",
       "      <td>141.91</td>\n",
       "      <td>34.07</td>\n",
       "      <td>35682.00</td>\n",
       "      <td>1243.33</td>\n",
       "      <td>41.95</td>\n",
       "      <td>-114.31</td>\n",
       "      <td>5.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         MedInc  HouseAge  AveRooms  AveBedrms  Population  AveOccup  \\\n",
       "count  20640.00  20640.00  20640.00   20436.00    20640.00  20640.00   \n",
       "mean       3.87     28.64      5.43       1.10     1425.48      3.07   \n",
       "std        1.90     12.59      2.47       0.48     1132.46     10.39   \n",
       "min        0.50      1.00      0.85       0.33        3.00      0.69   \n",
       "25%        2.56     18.00      4.44       1.01      787.00      2.43   \n",
       "50%        3.53     29.00      5.23       1.05     1166.00      2.82   \n",
       "75%        4.74     37.00      6.05       1.10     1725.00      3.28   \n",
       "max       15.00     52.00    141.91      34.07    35682.00   1243.33   \n",
       "\n",
       "       Latitude  Longitude  MedHouseVal  \n",
       "count  20640.00   20640.00     20640.00  \n",
       "mean      35.63    -119.57         2.07  \n",
       "std        2.14       2.00         1.15  \n",
       "min       32.54    -124.35         0.15  \n",
       "25%       33.93    -121.80         1.20  \n",
       "50%       34.26    -118.49         1.80  \n",
       "75%       37.71    -118.01         2.65  \n",
       "max       41.95    -114.31         5.00  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "housing.describe().round(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "39786ee8",
   "metadata": {},
   "source": [
    "> **Predict:** two columns are *capped*, meaning their max is an artificial ceiling rather than a real measurement. Which two, and how would you spot a cap in `describe()`? <details><summary>Answer</summary>`HouseAge` maxes at exactly 52.0 and `MedHouseVal` maxes at 5.00001 (i.e. \\$500,001). A cap shows up as a suspiciously round maximum with many rows piled exactly on it. The target cap matters most: the model can never predict above 5.0, so it will systematically under-predict the most expensive districts.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "20394df0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.377203Z",
     "iopub.status.busy": "2026-06-12T09:33:45.377117Z",
     "iopub.status.idle": "2026-06-12T09:33:45.379625Z",
     "shell.execute_reply": "2026-06-12T09:33:45.379259Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "HouseAge at the 52-year cap: 1273\n",
      "MedHouseVal at the $500k cap: 992\n"
     ]
    }
   ],
   "source": [
    "# how many districts sit exactly at each cap?\n",
    "at_age_cap = int((housing[\"HouseAge\"] >= 52.0).sum())\n",
    "at_val_cap = int((housing[\"MedHouseVal\"] >= 5.0).sum())\n",
    "print(f\"HouseAge at the 52-year cap: {at_age_cap}\")\n",
    "print(f\"MedHouseVal at the $500k cap: {at_val_cap}\")\n",
    "assert at_val_cap > 500, \"the target cap should trap many districts; if not, the data changed under us\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aa122bab",
   "metadata": {},
   "source": [
    "> **Interpretation.** Over 900 districts are pinned at the target cap. The model's residuals will fan out badly there, because the truth was clipped before we ever saw it. We will see exactly that fan-out in the residual plot in Part 5.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "487624d4",
   "metadata": {},
   "source": [
    "### 2.2 Distributions: the histograms\n",
    "\n",
    "A single `hist` grid is the fastest way to see skew, caps, and multi-modality at once.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "5b32763a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:45.380517Z",
     "iopub.status.busy": "2026-06-12T09:33:45.380427Z",
     "iopub.status.idle": "2026-06-12T09:33:46.003280Z",
     "shell.execute_reply": "2026-06-12T09:33:46.002808Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x800 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# viz: one histogram per numeric column\n",
    "num_only = housing.select_dtypes(include=\"number\")\n",
    "num_only.hist(bins=50, figsize=(12, 8))\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "011f9a58",
   "metadata": {},
   "source": [
    "> **What is the interpretation of this grid?** <details><summary>Answer</summary>The count-like and average columns (`Population`, `AveRooms`, `AveOccup`) are heavily right-skewed with long tails. `MedInc` is right-skewed and bounded. `HouseAge` and `MedHouseVal` show a spike at their caps. Skew and caps both argue for either a robust metric or a transform; we lead with the robust median AE (the framing's baseline metric) and keep RMSE as a tail-sensitive companion, rather than transforming the target, so the chapter's headline metric stays comparable to the \\$30k baseline.</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2783c8a",
   "metadata": {},
   "source": [
    "### 2.3 Geography: the scatter that draws California\n",
    "\n",
    "Plot the districts at their coordinates, sized by population and colored by value. The state appears, and so does the price gradient.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "de1ee934",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.004310Z",
     "iopub.status.busy": "2026-06-12T09:33:46.004225Z",
     "iopub.status.idle": "2026-06-12T09:33:46.228019Z",
     "shell.execute_reply": "2026-06-12T09:33:46.227629Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# viz: geographic scatter, color = value, size = population\n",
    "fig, ax = plt.subplots(figsize=(8, 6))\n",
    "sc = ax.scatter(housing[\"Longitude\"], housing[\"Latitude\"],\n",
    "                c=housing[\"MedHouseVal\"], cmap=\"viridis\",\n",
    "                s=housing[\"Population\"] / 80, alpha=0.35)\n",
    "ax.set_xlabel(\"longitude\"); ax.set_ylabel(\"latitude\")\n",
    "fig.colorbar(sc, ax=ax, label=\"median house value ($100k)\")\n",
    "plt.title(\"California districts: value by location\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "539b7bb6",
   "metadata": {},
   "source": [
    "> **Interpretation.** The coastline and the two big metros (the Bay Area in the north, Los Angeles in the south) light up in yellow. Geography carries real signal: coastal and metro districts are worth more. That is why latitude and longitude, weak on their own correlations, still earn their place as features (a tree can split on them jointly).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7404500e",
   "metadata": {},
   "source": [
    "### Exercise 2.1 — Rank features by correlation with the target\n",
    "\n",
    "`Difficulty 2/5 · ~8 min`\n",
    "\n",
    "Fill in `target_corrs(df, target)` returning a pandas `Series` of each numeric column's Pearson correlation with the target, sorted descending, with the target's own self-correlation (1.0) dropped. This is the single most informative line of the whole EDA: it tells you which feature carries the signal.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "9b258d6a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.229334Z",
     "iopub.status.busy": "2026-06-12T09:33:46.229237Z",
     "iopub.status.idle": "2026-06-12T09:33:46.232726Z",
     "shell.execute_reply": "2026-06-12T09:33:46.232433Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] 2.1 income dominates: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def target_corrs(df, target=\"MedHouseVal\"):\n",
    "    \"\"\"Pearson correlation of every numeric column with `target`,\n",
    "    sorted high-to-low, with the target's self-correlation removed.\n",
    "    Returns a pandas Series indexed by column name.\"\"\"\n",
    "    # TODO 1: compute the correlation matrix of the numeric columns (df.corr(numeric_only=True))\n",
    "    corr = None\n",
    "    # TODO 2: pull out the target's column/row, sort descending\n",
    "    series = None\n",
    "    # TODO 3: drop the target's own entry (its self-correlation is 1.0)\n",
    "    attempted(corr, series)\n",
    "    return series.drop(labels=[target])\n",
    "\n",
    "def _corr_signal():\n",
    "    s = target_corrs(housing)\n",
    "    # MedInc must dominate; it is ~+0.69 and everything else is well below it.\n",
    "    assert s.index[0] == \"MedInc\", f\"top feature was {s.index[0]!r}, expected 'MedInc'\"\n",
    "    assert s.iloc[0] > 0.6, f\"top correlation {s.iloc[0]:.2f} should exceed 0.6\"\n",
    "    assert \"MedHouseVal\" not in s.index, \"you left the target's self-correlation in\"\n",
    "\n",
    "check(\"2.1 income dominates\", _corr_signal)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3d835c9a",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>`df.corr(numeric_only=True)` gives a square matrix. Index it by the target name to get one column, then `.sort_values(ascending=False)`.</details>\n",
    "\n",
    "<details><summary>Hint 2 (pseudocode)</summary>\n",
    "\n",
    "```python\n",
    "corr = df.corr(numeric_only=True)\n",
    "series = corr[target].sort_values(ascending=False)\n",
    "# the function body already drops the target label for you\n",
    "```\n",
    "</details>\n",
    "\n",
    "<details><summary>Help — \"KeyError: 'MedHouseVal'\"</summary>You indexed the matrix before sorting, or you sorted the wrong axis. `corr[target]` selects the target's column; that column is a Series you can sort directly.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "a0edbf87",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.233812Z",
     "iopub.status.busy": "2026-06-12T09:33:46.233720Z",
     "iopub.status.idle": "2026-06-12T09:33:46.241837Z",
     "shell.execute_reply": "2026-06-12T09:33:46.241538Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 2.1 income dominates\n",
      "MedInc        0.688\n",
      "AveRooms      0.152\n",
      "HouseAge      0.106\n",
      "AveOccup     -0.024\n",
      "Population   -0.025\n",
      "Longitude    -0.046\n",
      "AveBedrms    -0.047\n",
      "Latitude     -0.144\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines target_corrs; the check below re-verifies the reference.\n",
    "def target_corrs(df, target=\"MedHouseVal\"):\n",
    "    corr = df.corr(numeric_only=True)\n",
    "    series = corr[target].sort_values(ascending=False)\n",
    "    return series.drop(labels=[target])\n",
    "\n",
    "check(\"2.1 income dominates\", _corr_signal, required=True)\n",
    "print(target_corrs(housing).round(3).to_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88db2fd1",
   "metadata": {},
   "source": [
    "> **Interpretation.** `MedInc` at about +0.69, then a steep drop. No other raw feature breaks 0.16. This is the diagnostic that tells you a linear model on income alone is a real baseline, and that the gains will come from features and models that can combine the weak signals (geography, occupancy) nonlinearly.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a62b8b1",
   "metadata": {},
   "source": [
    "### Exercise 2.2 — Engineer a ratio feature that beats its parents\n",
    "\n",
    "`Difficulty 2/5 · ~8 min`\n",
    "\n",
    "Raw counts are often weaker than ratios. Fill in `add_bedroom_ratio(df)` to add a `bedrooms_per_room = AveBedrms / AveRooms` column. Then we check that its correlation magnitude with the target exceeds `AveBedrms`'s own. Engineered features earning their keep is a theme that returns in Part 5's importances.\n",
    "\n",
    "Harder: also add `rooms_per_person = AveRooms / AveOccup` and print its correlation.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "47210517",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.242854Z",
     "iopub.status.busy": "2026-06-12T09:33:46.242776Z",
     "iopub.status.idle": "2026-06-12T09:33:46.246784Z",
     "shell.execute_reply": "2026-06-12T09:33:46.246323Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] 2.2 ratio beats parent: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def add_bedroom_ratio(df):\n",
    "    \"\"\"Return a copy of df with a new column bedrooms_per_room = AveBedrms / AveRooms.\"\"\"\n",
    "    out = df.copy()\n",
    "    # TODO 1: out[\"bedrooms_per_room\"] = AveBedrms divided by AveRooms\n",
    "    out[\"bedrooms_per_room\"] = None\n",
    "    attempted(out[\"bedrooms_per_room\"].iloc[0])\n",
    "    return out\n",
    "\n",
    "def _ratio_beats_parent():\n",
    "    enriched = add_bedroom_ratio(housing)\n",
    "    c = enriched.corr(numeric_only=True)[\"MedHouseVal\"]\n",
    "    ratio_strength = abs(c[\"bedrooms_per_room\"])\n",
    "    parent_strength = abs(c[\"AveBedrms\"])\n",
    "    assert ratio_strength > parent_strength, (\n",
    "        f\"bedrooms_per_room |corr|={ratio_strength:.3f} should beat \"\n",
    "        f\"AveBedrms |corr|={parent_strength:.3f}; check your division order\")\n",
    "\n",
    "check(\"2.2 ratio beats parent\", _ratio_beats_parent)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d67dc72",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>Two columns divide elementwise in pandas: `df[\"a\"] / df[\"b\"]` returns a new Series aligned by index.</details>\n",
    "\n",
    "<details><summary>Hint 2 (the line)</summary>`out[\"bedrooms_per_room\"] = out[\"AveBedrms\"] / out[\"AveRooms\"]`. Bedrooms on top, rooms on the bottom.</details>\n",
    "\n",
    "<details><summary>Help — the correlation came out NaN</summary>You divided by a column that contains zeros, or you left the NaNs from 1.3 in `AveBedrms`. pandas `.corr` drops NaN pairs, so a NaN result usually means a divide-by-zero produced `inf`. Check `AveRooms.min()`.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a50e5a8e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.247715Z",
     "iopub.status.busy": "2026-06-12T09:33:46.247637Z",
     "iopub.status.idle": "2026-06-12T09:33:46.258613Z",
     "shell.execute_reply": "2026-06-12T09:33:46.258235Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 2.2 ratio beats parent\n",
      "AveBedrms        corr = -0.047\n",
      "bedrooms_per_room corr = -0.255\n",
      "rooms_per_person  corr = +0.209\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines add_bedroom_ratio; the check below re-verifies the reference.\n",
    "def add_bedroom_ratio(df):\n",
    "    out = df.copy()\n",
    "    out[\"bedrooms_per_room\"] = out[\"AveBedrms\"] / out[\"AveRooms\"]\n",
    "    return out\n",
    "\n",
    "check(\"2.2 ratio beats parent\", _ratio_beats_parent, required=True)\n",
    "enriched = add_bedroom_ratio(housing)\n",
    "c = enriched.corr(numeric_only=True)[\"MedHouseVal\"]\n",
    "print(f\"AveBedrms        corr = {c['AveBedrms']:+.3f}\")\n",
    "print(f\"bedrooms_per_room corr = {c['bedrooms_per_room']:+.3f}\")\n",
    "# Harder: rooms per person\n",
    "print(f\"rooms_per_person  corr = {(enriched['AveRooms']/enriched['AveOccup']).corr(enriched['MedHouseVal']):+.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f5c52e6f",
   "metadata": {},
   "source": [
    "> **Interpretation.** The ratio carries more signal than its parent count, which is the whole argument for feature engineering. We will fold this kind of ratio into the pipeline as a custom transformer in Part 4 so it is computed the same way at train time and at inference time.\n",
    "\n",
    "> **Key takeaways**\n",
    "> - `info` and `describe` first; caps and gaps live in the ranges.\n",
    "> - Histograms reveal skew; the geographic scatter reveals that location is signal.\n",
    "> - `MedInc` dominates the correlation ranking; engineered ratios can beat raw counts.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d8b4ecb",
   "metadata": {},
   "source": [
    "## Part 3 — The split is a decision\n",
    "\n",
    "> **Objectives**\n",
    "> - Bin income into strata and split so the strata are preserved.\n",
    "> - Quantify how a random split biases the income mix versus a stratified one.\n",
    "> - Lock the test set away and not look at it again until Part 6.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4010973d",
   "metadata": {},
   "source": [
    "### 3.1 Why stratify on income\n",
    "\n",
    "A random split is fine when the data is large and well mixed. But income is so predictive here that if the train and test halves end up with different income mixes, the test score measures the mix as much as the model. The fix is to bin income into a few categories and sample each category proportionally into both halves. We bin with the same edges Géron uses, chosen so each bucket has enough districts to be stable.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "3319ff37",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.259685Z",
     "iopub.status.busy": "2026-06-12T09:33:46.259605Z",
     "iopub.status.idle": "2026-06-12T09:33:46.263207Z",
     "shell.execute_reply": "2026-06-12T09:33:46.262847Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "income_cat\n",
      "1     822\n",
      "2    6581\n",
      "3    7236\n",
      "4    3639\n",
      "5    2362\n"
     ]
    }
   ],
   "source": [
    "# income strata: edges chosen so each bucket is populated (Géron's bins)\n",
    "income_bins = [0.0, 1.5, 3.0, 4.5, 6.0, np.inf]\n",
    "housing[\"income_cat\"] = pd.cut(housing[\"MedInc\"], bins=income_bins, labels=[1, 2, 3, 4, 5])\n",
    "print(housing[\"income_cat\"].value_counts().sort_index().to_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "71aa4b23",
   "metadata": {},
   "source": [
    "> **Interpretation.** Five strata, with the middle buckets fattest and the extremes thin. The thin extreme buckets are exactly the ones a random split can under-sample by chance, biasing the evaluation.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1e29b163",
   "metadata": {},
   "source": [
    "### Exercise 3.1 — A stratified split that preserves the income mix\n",
    "\n",
    "`Difficulty 3/5 · ~15 min`\n",
    "\n",
    "Fill in `make_split(df, strat_col, test_size, seed)` using `StratifiedShuffleSplit` to return `(train_df, test_df)`, with the helper `strat_col` dropped from both. The seed is fixed so the checks know what to expect, the same reproducibility discipline you used in Ch 00. We then verify the test set's stratum proportions match the full data to within a tight tolerance.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "816a15fe",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.264075Z",
     "iopub.status.busy": "2026-06-12T09:33:46.264002Z",
     "iopub.status.idle": "2026-06-12T09:33:46.280262Z",
     "shell.execute_reply": "2026-06-12T09:33:46.279793Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] 3.1 stratified split: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import StratifiedShuffleSplit\n",
    "\n",
    "def make_split(df, strat_col, test_size=0.2, seed=SEED):\n",
    "    \"\"\"Stratified train/test split. Returns (train_df, test_df) with strat_col dropped.\"\"\"\n",
    "    # TODO 1: build StratifiedShuffleSplit(n_splits=1, test_size=test_size, random_state=seed)\n",
    "    splitter = None\n",
    "    # TODO 2: get the single (train_idx, test_idx) pair: next(splitter.split(df, df[strat_col]))\n",
    "    train_idx, test_idx = None, None\n",
    "    attempted(splitter, train_idx, test_idx)\n",
    "    train_df = df.iloc[train_idx].drop(columns=strat_col)\n",
    "    test_df = df.iloc[test_idx].drop(columns=strat_col)\n",
    "    return train_df, test_df\n",
    "\n",
    "def _strat_preserved():\n",
    "    tr, te = make_split(housing, \"income_cat\", test_size=0.2, seed=SEED)\n",
    "    # recompute strata on the test split and compare proportions to the full data\n",
    "    full_p = housing[\"income_cat\"].value_counts(normalize=True).sort_index()\n",
    "    te_cat = pd.cut(te[\"MedInc\"], bins=income_bins, labels=[1, 2, 3, 4, 5])\n",
    "    te_p = te_cat.value_counts(normalize=True).sort_index()\n",
    "    max_diff = float((full_p - te_p).abs().max())\n",
    "    assert max_diff < 0.005, (\n",
    "        f\"max stratum-proportion drift {max_diff:.4f} is too large; \"\n",
    "        \"a stratified split should keep it near zero (a random split would be ~10x worse)\")\n",
    "    assert \"income_cat\" not in tr.columns, \"you forgot to drop the income_cat helper column\"\n",
    "\n",
    "check(\"3.1 stratified split\", _strat_preserved)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c876d60",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>`StratifiedShuffleSplit` yields index arrays, not data. You give `.split` the frame and the stratification column, and it hands back `(train_idx, test_idx)`. With `n_splits=1` there is exactly one pair, so `next(...)` is the idiom.</details>\n",
    "\n",
    "<details><summary>Hint 2 (pseudocode)</summary>\n",
    "\n",
    "```python\n",
    "splitter = StratifiedShuffleSplit(n_splits=1, test_size=test_size, random_state=seed)\n",
    "train_idx, test_idx = next(splitter.split(df, df[strat_col]))\n",
    "# the function body already does the .iloc and the column drop\n",
    "```\n",
    "</details>\n",
    "\n",
    "<details><summary>Help — \"Found input variables with inconsistent numbers of samples\"</summary>You passed the stratification column as a 2D frame instead of a 1D Series, or you passed `df` twice. The first argument is the data to index, the second is the 1D strata: `splitter.split(df, df[strat_col])`.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b3a66557",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.281170Z",
     "iopub.status.busy": "2026-06-12T09:33:46.281090Z",
     "iopub.status.idle": "2026-06-12T09:33:46.295451Z",
     "shell.execute_reply": "2026-06-12T09:33:46.295065Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 3.1 stratified split\n",
      "train (16512, 10) · test (4128, 10)\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines make_split; the check below re-verifies the reference.\n",
    "def make_split(df, strat_col, test_size=0.2, seed=SEED):\n",
    "    splitter = StratifiedShuffleSplit(n_splits=1, test_size=test_size, random_state=seed)\n",
    "    train_idx, test_idx = next(splitter.split(df, df[strat_col]))\n",
    "    train_df = df.iloc[train_idx].drop(columns=strat_col)\n",
    "    test_df = df.iloc[test_idx].drop(columns=strat_col)\n",
    "    return train_df, test_df\n",
    "\n",
    "check(\"3.1 stratified split\", _strat_preserved, required=True)\n",
    "strat_train, strat_test = make_split(housing, \"income_cat\", test_size=0.2, seed=SEED)\n",
    "print(f\"train {strat_train.shape} · test {strat_test.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d70f4111",
   "metadata": {},
   "source": [
    "### 3.2 How biased would a random split have been?\n",
    "\n",
    "Claim from the prose: a random split skews the income mix more than a stratified one. Let us not take that on faith. We compute the per-stratum proportion error for both kinds of split and compare. This is the spanish-inquisition rule: a claim that can be an assert should be one.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "7764327d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.296507Z",
     "iopub.status.busy": "2026-06-12T09:33:46.296433Z",
     "iopub.status.idle": "2026-06-12T09:33:46.303296Z",
     "shell.execute_reply": "2026-06-12T09:33:46.302942Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stratified split total stratum error: 0.0004\n",
      "random split     total stratum error: 0.0143\n",
      "the random split is 37.0x more biased on the income mix\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "def stratum_error(test_subset):\n",
    "    full_p = housing[\"income_cat\"].value_counts(normalize=True).sort_index()\n",
    "    cats = pd.cut(test_subset[\"MedInc\"], bins=income_bins, labels=[1, 2, 3, 4, 5])\n",
    "    te_p = cats.value_counts(normalize=True).sort_index()\n",
    "    return float((full_p - te_p).abs().sum())   # total absolute proportion error\n",
    "\n",
    "# random split (same size, same seed) vs the stratified test set from 3.1\n",
    "rand_train, rand_test = train_test_split(housing, test_size=0.2, random_state=SEED)\n",
    "strat_err = stratum_error(strat_test)\n",
    "rand_err = stratum_error(rand_test)\n",
    "print(f\"stratified split total stratum error: {strat_err:.4f}\")\n",
    "print(f\"random split     total stratum error: {rand_err:.4f}\")\n",
    "assert strat_err < rand_err, \"stratifying should reduce stratum drift; if not, check the bins\"\n",
    "print(f\"the random split is {rand_err / max(strat_err, 1e-9):.1f}x more biased on the income mix\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "61b18fbc",
   "metadata": {},
   "source": [
    "> **Interpretation.** The random split distorts the income mix several times more than the stratified one. On a small test set that distortion becomes test-score noise you cannot diagnose. The general rule: if a feature is predictive enough that imbalance in it would bias evaluation, stratify on a binned version of it. Time series get a different rule entirely, a chronological split, never random.\n",
    "\n",
    "> **Caveat:** drop the `income_cat` helper before modeling (the split already did). It was scaffolding for the split, not a feature. Leaving engineered split-helpers in the feature matrix is a quiet way to leak the very structure you split on.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "308c9bda",
   "metadata": {},
   "source": [
    "> **Key takeaways**\n",
    "> - Stratify on a binned version of any feature predictive enough to bias evaluation.\n",
    "> - We proved the random split was several times more biased, rather than asserting it.\n",
    "> - The test set (`strat_test`) is now sealed; we do not touch it until Part 6.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "348ec12d",
   "metadata": {},
   "source": [
    "## Part 4 — The pipeline is the leakage firewall\n",
    "\n",
    "> **Objectives**\n",
    "> - Separate features from target, and name the numeric vs categorical columns.\n",
    "> - Build a custom transformer for the engineered ratios that runs inside the pipeline.\n",
    "> - Compose imputation, scaling, encoding into one `ColumnTransformer` that cannot leak.\n",
    "> - Watch a leak fabricate a score on a noise target, then close it structurally.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0257a6a9",
   "metadata": {},
   "source": [
    "### 4.1 Separate features and target\n",
    "\n",
    "The target never goes through the feature pipeline. We split it off first, on both halves.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "61ca28de",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.304171Z",
     "iopub.status.busy": "2026-06-12T09:33:46.304086Z",
     "iopub.status.idle": "2026-06-12T09:33:46.307269Z",
     "shell.execute_reply": "2026-06-12T09:33:46.306898Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train (16512, 9) · numeric 8 · categorical 1\n"
     ]
    }
   ],
   "source": [
    "X_train = strat_train.drop(columns=\"MedHouseVal\")\n",
    "y_train = strat_train[\"MedHouseVal\"].copy()\n",
    "X_test = strat_test.drop(columns=\"MedHouseVal\")\n",
    "y_test = strat_test[\"MedHouseVal\"].copy()\n",
    "\n",
    "num_columns = [\"MedInc\", \"HouseAge\", \"AveRooms\", \"AveBedrms\",\n",
    "               \"Population\", \"AveOccup\", \"Latitude\", \"Longitude\"]\n",
    "cat_columns = [\"ocean_proximity\"]\n",
    "print(f\"X_train {X_train.shape} · numeric {len(num_columns)} · categorical {len(cat_columns)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2a6cad5",
   "metadata": {},
   "source": [
    "### Exercise 4.1 — A custom transformer for the ratio feature\n",
    "\n",
    "`Difficulty 3/5 · ~15 min`\n",
    "\n",
    "A pipeline stage is any object with `fit` and `transform`. To compute the `bedrooms_per_room` ratio the same way at train and inference time, wrap it in a transformer. Fill in `RatioAdder.transform` to append the ratio column. The transformer receives a NumPy array (the `ColumnTransformer` passes arrays), and `bedrooms_idx` / `rooms_idx` tell it which columns to divide.\n",
    "\n",
    "The micro-demo below feeds a 2-row toy array with hand-checkable values, so you can verify the output before touching real data.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "011231d5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.308249Z",
     "iopub.status.busy": "2026-06-12T09:33:46.308169Z",
     "iopub.status.idle": "2026-06-12T09:33:46.312565Z",
     "shell.execute_reply": "2026-06-12T09:33:46.312323Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] 4.1 RatioAdder toy: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "\n",
    "class RatioAdder(BaseEstimator, TransformerMixin):\n",
    "    \"\"\"Append AveBedrms / AveRooms as a new last column.\"\"\"\n",
    "    def __init__(self, bedrooms_idx=3, rooms_idx=2):\n",
    "        self.bedrooms_idx = bedrooms_idx\n",
    "        self.rooms_idx = rooms_idx\n",
    "    def fit(self, X, y=None):\n",
    "        return self  # stateless: nothing to learn, so nothing to leak\n",
    "    def transform(self, X):\n",
    "        X = np.asarray(X, dtype=float)\n",
    "        # TODO 1: ratio = column bedrooms_idx divided by column rooms_idx\n",
    "        ratio = None\n",
    "        attempted(ratio)\n",
    "        # TODO 2: stack ratio as a new last column (np.c_[X, ratio])\n",
    "        return np.c_[X, ratio]\n",
    "\n",
    "def _ratio_adder_toy():\n",
    "    # toy: 2 rows, 4 cols [rooms_idx=2, bedrooms_idx=3] -> ratio = col3/col2\n",
    "    toy = np.array([[0., 0., 4., 1.],     # 1/4 = 0.25\n",
    "                    [0., 0., 5., 2.]])     # 2/5 = 0.40\n",
    "    out = RatioAdder(bedrooms_idx=3, rooms_idx=2).fit_transform(toy)\n",
    "    assert out.shape == (2, 5), f\"expected one new column -> shape (2,5), got {out.shape}\"\n",
    "    check_close(out[:, -1], [0.25, 0.40], msg=\"ratio = bedrooms/rooms: 1/4 and 2/5\")\n",
    "\n",
    "check(\"4.1 RatioAdder toy\", _ratio_adder_toy)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bfe19581",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>Index a 2D NumPy array's column with `X[:, j]`. The ratio is `X[:, bedrooms_idx] / X[:, rooms_idx]`. `np.c_[X, ratio]` glues a column on the right.</details>\n",
    "\n",
    "<details><summary>Hint 2 (pseudocode)</summary>\n",
    "\n",
    "```python\n",
    "ratio = X[:, self.bedrooms_idx] / X[:, self.rooms_idx]\n",
    "return np.c_[X, ratio]\n",
    "```\n",
    "</details>\n",
    "\n",
    "<details><summary>Help — \"could not broadcast\" or the shape is wrong</summary>`np.c_[X, ratio]` needs `ratio` to be 1D of length `n_rows`. If you wrote `X[:, j:j+1]` you got a 2D column; either is fine for `np.c_`, but mixing a 2D `X` with a 0D scalar ratio (from using the wrong index) fails. Print `ratio.shape`.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "d596b1a6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.313423Z",
     "iopub.status.busy": "2026-06-12T09:33:46.313352Z",
     "iopub.status.idle": "2026-06-12T09:33:46.316083Z",
     "shell.execute_reply": "2026-06-12T09:33:46.315612Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 4.1 RatioAdder toy\n",
      "RatioAdder verified on the toy input\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines RatioAdder.transform; the check below re-verifies the reference.\n",
    "class RatioAdder(BaseEstimator, TransformerMixin):\n",
    "    def __init__(self, bedrooms_idx=3, rooms_idx=2):\n",
    "        self.bedrooms_idx = bedrooms_idx\n",
    "        self.rooms_idx = rooms_idx\n",
    "    def fit(self, X, y=None):\n",
    "        return self\n",
    "    def transform(self, X):\n",
    "        X = np.asarray(X, dtype=float)\n",
    "        ratio = X[:, self.bedrooms_idx] / X[:, self.rooms_idx]\n",
    "        return np.c_[X, ratio]\n",
    "\n",
    "check(\"4.1 RatioAdder toy\", _ratio_adder_toy, required=True)\n",
    "print(\"RatioAdder verified on the toy input\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0004312b",
   "metadata": {},
   "source": [
    "> **Note:** `RatioAdder` is stateless, its `fit` learns nothing, so it cannot leak. The stages that *do* learn (the imputer's median, the scaler's mean and std) are the dangerous ones, and the pipeline is what guarantees they only ever see training rows.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6e701557",
   "metadata": {},
   "source": [
    "### 4.2 Compose the firewall\n",
    "\n",
    "The numeric branch: impute missing values with the median, append the ratio, then scale. The categorical branch: one-hot encode `ocean_proximity`. A `ColumnTransformer` routes each set of columns to its branch and concatenates the results. The whole thing exposes one `fit_transform` (for train) and one `transform` (for test). You fit on train, transform test. You cannot accidentally do otherwise.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "82b9dfae",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.316892Z",
     "iopub.status.busy": "2026-06-12T09:33:46.316820Z",
     "iopub.status.idle": "2026-06-12T09:33:46.356256Z",
     "shell.execute_reply": "2026-06-12T09:33:46.355734Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "prepared train (16512, 13) · prepared test (4128, 13)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.preprocessing import StandardScaler, OneHotEncoder\n",
    "\n",
    "num_pipeline = Pipeline([\n",
    "    (\"impute\", SimpleImputer(strategy=\"median\")),   # learns the median from TRAIN only\n",
    "    (\"ratios\", RatioAdder()),                        # stateless ratio feature\n",
    "    (\"scale\", StandardScaler()),                     # learns mean/std from TRAIN only\n",
    "])\n",
    "\n",
    "full_pipeline = ColumnTransformer([\n",
    "    (\"num\", num_pipeline, num_columns),\n",
    "    (\"cat\", OneHotEncoder(handle_unknown=\"ignore\"), cat_columns),  # unseen categories -> all-zero, no crash\n",
    "])\n",
    "\n",
    "X_train_prepared = full_pipeline.fit_transform(X_train)   # fit_transform on TRAIN\n",
    "X_test_prepared = full_pipeline.transform(X_test)         # transform (not fit) on TEST\n",
    "print(f\"prepared train {X_train_prepared.shape} · prepared test {X_test_prepared.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "82db0d60",
   "metadata": {},
   "source": [
    "> **Interpretation.** Eight numeric columns plus one engineered ratio (nine) plus four one-hot categories gives thirteen prepared columns. The train and test column counts match, which is the first thing to check after any `ColumnTransformer`: a mismatch means a category appeared in one half and not the other, and `handle_unknown=\"ignore\"` is what keeps that from crashing inference.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "5814872b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.357369Z",
     "iopub.status.busy": "2026-06-12T09:33:46.357269Z",
     "iopub.status.idle": "2026-06-12T09:33:46.359908Z",
     "shell.execute_reply": "2026-06-12T09:33:46.359385Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "imputer medians (per numeric column), learned from train:\n",
      "[ 3.5340e+00  2.9000e+01  5.2260e+00  1.0490e+00  1.1670e+03  2.8220e+00\n",
      "  3.4250e+01 -1.1849e+02]\n"
     ]
    }
   ],
   "source": [
    "# the imputer's learned median came from train only; confirm it is a fitted statistic, not a global one\n",
    "learned_median = full_pipeline.named_transformers_[\"num\"].named_steps[\"impute\"].statistics_\n",
    "print(\"imputer medians (per numeric column), learned from train:\")\n",
    "print(np.round(learned_median, 3))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56df41e2",
   "metadata": {},
   "source": [
    "> **Notice that** the imputer stored one median per numeric column, and those medians were computed on the training rows alone. When `transform` runs on the test set, it reuses these stored medians; it never recomputes them from test data. That is the leak, sealed.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a8f6c39",
   "metadata": {},
   "source": [
    "### 4.3 A deliberate failure: leakage fabricates a score\n",
    "\n",
    "Here is the bug the firewall prevents, staged so you feel it. Take a target that is **pure noise**, unrelated to any feature. A clean evaluation must report no predictive power. But if you select the \"best\" features using the whole dataset *before* cross-validating, the selection peeks at the held-out folds and fabricates a score. We run the broken version first, see the lie, then fix it with one structural change.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "e4a07fc0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.360725Z",
     "iopub.status.busy": "2026-06-12T09:33:46.360651Z",
     "iopub.status.idle": "2026-06-12T09:33:46.373335Z",
     "shell.execute_reply": "2026-06-12T09:33:46.372739Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[broken] cross-validated R^2 after leaky feature selection: 0.405\n"
     ]
    }
   ],
   "source": [
    "# scale-up: a small noise dataset where the TRUE signal is zero, so any positive score is fake\n",
    "from sklearn.feature_selection import SelectKBest, f_regression\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "rng_leak = np.random.default_rng(SEED)           # local RNG so this demo never perturbs the main run\n",
    "n_demo, p_demo = 100, 2000\n",
    "X_noise = rng_leak.normal(size=(n_demo, p_demo))\n",
    "y_noise = rng_leak.normal(size=n_demo)           # target is independent noise: true R^2 is ~0\n",
    "\n",
    "# BROKEN: pick the 20 features most correlated with y using ALL the data, THEN cross-validate\n",
    "selector = SelectKBest(f_regression, k=20).fit(X_noise, y_noise)   # <- peeks at every row, including holdout folds\n",
    "X_selected = selector.transform(X_noise)\n",
    "leaked_r2 = cross_val_score(LinearRegression(), X_selected, y_noise, scoring=\"r2\", cv=5).mean()\n",
    "print(f\"[broken] cross-validated R^2 after leaky feature selection: {leaked_r2:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17f1831f",
   "metadata": {},
   "source": [
    "> **Interpretation.** A model predicting pure noise just reported a clearly positive R². It is entirely fake: the feature selection looked at the same rows the cross-validation later \"held out\", so the chosen features were tuned to those folds. This is precisely how a leaked test score lies, and why a test score that looks *too good* is an alarm, not a celebration.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "090fe8d9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.374248Z",
     "iopub.status.busy": "2026-06-12T09:33:46.374104Z",
     "iopub.status.idle": "2026-06-12T09:33:46.388495Z",
     "shell.execute_reply": "2026-06-12T09:33:46.388001Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[fixed]  cross-validated R^2 with selection inside the pipeline: -0.133\n",
      "the honest score is near zero (often negative): correct, because the target was noise\n"
     ]
    }
   ],
   "source": [
    "# FIXED: put the selection INSIDE the pipeline, so it refits on each CV training fold only\n",
    "clean_pipe = Pipeline([\n",
    "    (\"select\", SelectKBest(f_regression, k=20)),   # now fit only on each fold's training rows\n",
    "    (\"lr\", LinearRegression()),\n",
    "])\n",
    "clean_r2 = cross_val_score(clean_pipe, X_noise, y_noise, scoring=\"r2\", cv=5).mean()\n",
    "print(f\"[fixed]  cross-validated R^2 with selection inside the pipeline: {clean_r2:.3f}\")\n",
    "assert clean_r2 < leaked_r2, \"the fixed pipeline must report a worse (honest) score than the leaky one\"\n",
    "print(\"the honest score is near zero (often negative): correct, because the target was noise\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c69f18d4",
   "metadata": {},
   "source": [
    "> **Interpretation.** The fix is not a flag or a parameter; it is structure. Once the selection lives inside the pipeline, cross-validation refits it on each fold's training rows alone, and the score collapses to honest near-zero. Every preprocessing step that *learns* anything (a median, a mean, a vocabulary, a feature ranking, a PCA basis) must be a pipeline stage. That is the entire reason `Pipeline` + `ColumnTransformer` exists.\n",
    "\n",
    "> **Common confusion:** \"but my real pipeline above used `fit_transform` on the whole training set, isn't that the same leak?\" No. The training set is allowed to see itself. The leak is when the *test* (or CV holdout) influences a fitted statistic. Fitting on all of `X_train` is fine; the firewall is that `X_test` only ever meets `transform`.\n",
    "\n",
    "> **Key takeaways**\n",
    "> - Any step that learns a statistic must be a pipeline stage, or it can leak.\n",
    "> - The `ColumnTransformer` routes columns to branches and concatenates; `fit_transform` train, `transform` test.\n",
    "> - We watched leakage fabricate R² on a noise target, then closed it with one structural move.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "98995eb6",
   "metadata": {},
   "source": [
    "## Part 5 — Select, tune, and analyze\n",
    "\n",
    "> **Objectives**\n",
    "> - Compare three model families on one cross-validation harness.\n",
    "> - Tune the winner with randomized search and read its feature importances.\n",
    "> - Plot residuals and find where the model fails (the cap, the tails).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "faf6e92c",
   "metadata": {},
   "source": [
    "### 5.1 One harness, three model families\n",
    "\n",
    "Wrap the preprocessing and a model into one estimator so cross-validation re-runs the whole pipeline on each fold (no leak). Score with negative RMSE so larger is better in sklearn's convention, then flip the sign for reporting. We compare linear regression, a single decision tree, and a random forest.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "5ee59a47",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:46.389839Z",
     "iopub.status.busy": "2026-06-12T09:33:46.389701Z",
     "iopub.status.idle": "2026-06-12T09:33:48.564118Z",
     "shell.execute_reply": "2026-06-12T09:33:48.563762Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "LinearRegression   CV RMSE = 0.716 ± 0.009\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DecisionTree       CV RMSE = 0.759 ± 0.005\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RandomForest       CV RMSE = 0.522 ± 0.002\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "CV = 3                                  # 3-fold keeps the chapter fast; 5 is the usual default\n",
    "N_TREES = 30 if FAST else 80            # forest size: smaller in CI smoke mode, same code path\n",
    "\n",
    "def cv_rmse(model, X, y, cv=CV):\n",
    "    \"\"\"Mean and std of RMSE across cv folds (sklearn scores negative RMSE, we flip it).\"\"\"\n",
    "    scores = cross_val_score(model, X, y, scoring=\"neg_root_mean_squared_error\", cv=cv)\n",
    "    rmse = -scores\n",
    "    return rmse.mean(), rmse.std()\n",
    "\n",
    "candidates = {\n",
    "    \"LinearRegression\": LinearRegression(),\n",
    "    \"DecisionTree\": DecisionTreeRegressor(random_state=SEED),\n",
    "    \"RandomForest\": RandomForestRegressor(n_estimators=N_TREES, random_state=SEED, n_jobs=-1),\n",
    "}\n",
    "cv_results = {}\n",
    "for name, model in candidates.items():\n",
    "    est = Pipeline([(\"prep\", full_pipeline), (\"model\", model)])\n",
    "    m, s = cv_rmse(est, X_train, y_train)\n",
    "    cv_results[name] = m\n",
    "    print(f\"{name:18s} CV RMSE = {m:.3f} ± {s:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48047947",
   "metadata": {},
   "source": [
    "> **Predict:** which family wins, and why is the single decision tree the worst? <details><summary>Answer</summary>The random forest wins (RMSE ~0.52 vs ~0.72 linear, ~0.76 single tree). The single tree overfits: it can memorize the training folds (near-perfect train fit) but generalizes worse than even the linear model. The forest averages many decorrelated trees, cancelling that variance. This is the chapter's central lesson: the gap between model *families* (here ~0.24 RMSE) dwarfs the gap tuning will buy (a few hundredths). Pick the family before you tune.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "32e455a5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:48.565219Z",
     "iopub.status.busy": "2026-06-12T09:33:48.565074Z",
     "iopub.status.idle": "2026-06-12T09:33:48.567373Z",
     "shell.execute_reply": "2026-06-12T09:33:48.566988Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best family: RandomForest at CV RMSE 0.522\n"
     ]
    }
   ],
   "source": [
    "# assert the lesson rather than asserting it in prose\n",
    "assert cv_results[\"RandomForest\"] < cv_results[\"LinearRegression\"], \"forest should beat linear here\"\n",
    "assert cv_results[\"RandomForest\"] < cv_results[\"DecisionTree\"], \"forest should beat the single tree\"\n",
    "best_family = min(cv_results, key=cv_results.get)\n",
    "print(f\"best family: {best_family} at CV RMSE {cv_results[best_family]:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ac675a62",
   "metadata": {},
   "source": [
    "### Exercise 5.1 — The naive predictor (a secondary sanity floor)\n",
    "\n",
    "`Difficulty 1/5 · ~5 min`\n",
    "\n",
    "Two baselines live in this chapter, and it pays to keep them straight. The bar that *matters* is the framing's human-appraiser **median AE of \\$30k** (0.30 in \\$100k units); we compare the model's median AE against it when we touch the test set in Part 6. This exercise builds the *secondary, naive* floor: the RMSE you get by predicting the training mean for every row. It is the dumbest model that still respects the data — a sanity check, not the baseline to beat. Fill in `constant_baseline_rmse(y_train, y_eval)`: the RMSE of always guessing the training mean on `y_eval`.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "78eb68f7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:48.568180Z",
     "iopub.status.busy": "2026-06-12T09:33:48.568106Z",
     "iopub.status.idle": "2026-06-12T09:33:48.571679Z",
     "shell.execute_reply": "2026-06-12T09:33:48.571401Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] 5.1 constant baseline: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def constant_baseline_rmse(y_train, y_eval):\n",
    "    \"\"\"RMSE of always predicting the TRAIN mean on y_eval (a numpy array or Series).\"\"\"\n",
    "    y_train = np.asarray(y_train, dtype=float)\n",
    "    y_eval = np.asarray(y_eval, dtype=float)\n",
    "    # TODO 1: guess = the mean of y_train (a single number)\n",
    "    guess = None\n",
    "    attempted(guess)\n",
    "    # TODO 2: rmse = sqrt(mean((y_eval - guess)**2))\n",
    "    return float(np.sqrt(np.mean((y_eval - guess) ** 2)))\n",
    "\n",
    "def _baseline_sane():\n",
    "    # on the training set, the constant baseline RMSE equals the train std (population form)\n",
    "    b = constant_baseline_rmse(y_train, y_train)\n",
    "    expected = float(np.std(np.asarray(y_train, dtype=float)))\n",
    "    check_close(b, expected, atol=1e-6, msg=\"constant-mean RMSE on train == population std of train\")\n",
    "    # and the random forest must beat it comfortably\n",
    "    assert cv_results[\"RandomForest\"] < b, (\n",
    "        f\"forest CV RMSE {cv_results['RandomForest']:.3f} should beat the baseline {b:.3f}\")\n",
    "\n",
    "check(\"5.1 constant baseline\", _baseline_sane)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db7f5fcc",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>The constant that minimizes squared error is the mean. Compute it on train, broadcast it against every eval target, square the gaps, average, square-root.</details>\n",
    "\n",
    "<details><summary>Hint 2 (the line)</summary>`guess = y_train.mean()`. The function already does the RMSE with that guess.</details>\n",
    "\n",
    "<details><summary>Help — \"the train-set RMSE didn't equal the std\"</summary>RMSE of the mean equals the *population* std (divide by N), which is `np.std(...)` with default `ddof=0`. pandas `.std()` uses `ddof=1` (divide by N-1) and will be slightly larger. The check uses `np.std`, so use the mean of `y_train`, not its std, as your guess.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "e074b06b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:48.572494Z",
     "iopub.status.busy": "2026-06-12T09:33:48.572423Z",
     "iopub.status.idle": "2026-06-12T09:33:48.575396Z",
     "shell.execute_reply": "2026-06-12T09:33:48.575081Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 5.1 constant baseline\n",
      "naive predictor RMSE (guess train mean): 1.155   <- secondary sanity floor\n",
      "random forest CV RMSE:                   0.522  (55% better than guessing the mean)\n",
      "the bar that MATTERS is the $30k human-appraiser MEDIAN AE; we test it in Part 6\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines constant_baseline_rmse; the check below re-verifies the reference.\n",
    "def constant_baseline_rmse(y_train, y_eval):\n",
    "    y_train = np.asarray(y_train, dtype=float)\n",
    "    y_eval = np.asarray(y_eval, dtype=float)\n",
    "    guess = y_train.mean()\n",
    "    return float(np.sqrt(np.mean((y_eval - guess) ** 2)))\n",
    "\n",
    "check(\"5.1 constant baseline\", _baseline_sane, required=True)\n",
    "base = constant_baseline_rmse(y_train, y_train)\n",
    "print(f\"naive predictor RMSE (guess train mean): {base:.3f}   <- secondary sanity floor\")\n",
    "print(f\"random forest CV RMSE:                   {cv_results['RandomForest']:.3f}  \"\n",
    "      f\"({(1 - cv_results['RandomForest']/base):.0%} better than guessing the mean)\")\n",
    "print(\"the bar that MATTERS is the $30k human-appraiser MEDIAN AE; we test it in Part 6\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3832478",
   "metadata": {},
   "source": [
    "> **Interpretation.** The forest cuts the error roughly in half versus the naive mean-guess (RMSE ~0.52 vs ~1.15). Clearing the naive floor is necessary but cheap — it only proves the model learned *something*. The real bar is the human-appraiser **median AE of \\$30k** from the framing, and we compare the model's median AE (not its RMSE) against it once, on the test set, in Part 6. Keep the two baselines distinct: the mean-guess RMSE is the naive sanity floor; the \\$30k median AE is the baseline to beat.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0846c245",
   "metadata": {},
   "source": [
    "### 5.2 Tune the winner with randomized search\n",
    "\n",
    "Grid search enumerates every combination; randomized search samples from distributions and is more efficient when you do not know the right ranges. We wrap the full pipeline so each search candidate is cross-validated end to end (preprocessing included, no leak). The double-underscore in `model__n_estimators` reaches into the `model` step of the pipeline.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e815b612",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:48.576217Z",
     "iopub.status.busy": "2026-06-12T09:33:48.576139Z",
     "iopub.status.idle": "2026-06-12T09:33:56.030038Z",
     "shell.execute_reply": "2026-06-12T09:33:56.029733Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best params: {'model__max_features': 5, 'model__n_estimators': 107}\n",
      "best CV RMSE: 0.497  (untuned forest was 0.522)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from scipy.stats import randint\n",
    "\n",
    "N_ITER = 4 if FAST else 8               # search budget: tiny in CI, modest otherwise\n",
    "param_dist = {\n",
    "    \"model__n_estimators\": randint(40, 120),   # number of trees\n",
    "    \"model__max_features\": randint(2, 8),       # features considered per split (decorrelates trees)\n",
    "}\n",
    "rf_pipe = Pipeline([(\"prep\", full_pipeline),\n",
    "                    (\"model\", RandomForestRegressor(random_state=SEED, n_jobs=-1))])\n",
    "search = RandomizedSearchCV(\n",
    "    rf_pipe, param_dist, n_iter=N_ITER, cv=CV,\n",
    "    scoring=\"neg_root_mean_squared_error\", random_state=SEED, n_jobs=-1)\n",
    "search.fit(X_train, y_train)\n",
    "print(f\"best params: {search.best_params_}\")\n",
    "print(f\"best CV RMSE: {-search.best_score_:.3f}  (untuned forest was {cv_results['RandomForest']:.3f})\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0db55838",
   "metadata": {},
   "source": [
    "> **Interpretation.** Tuning shaves a few hundredths off the RMSE. Real, but small, and far smaller than the ~0.24 the random forest already won over linear regression by being the right family. Internalize the ratio: most practitioners over-invest in hyperparameter search and under-invest in feature engineering and model selection.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9bda03db",
   "metadata": {},
   "source": [
    "### Exercise 5.2 — Read the feature importances\n",
    "\n",
    "`Difficulty 2/5 · ~10 min`\n",
    "\n",
    "A fitted forest exposes `feature_importances_`, one number per *prepared* column. Fill in `top_importances(search, feature_names, k)` to pair the importances with their names and return the top `k` as a list of `(name, importance)` tuples, highest first. We then verify income is the single most important feature, matching the EDA.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "3361fdf7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.031058Z",
     "iopub.status.busy": "2026-06-12T09:33:56.030977Z",
     "iopub.status.idle": "2026-06-12T09:33:56.048191Z",
     "shell.execute_reply": "2026-06-12T09:33:56.047839Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "13 prepared features: ['MedInc', 'HouseAge', 'AveRooms', 'AveBedrms', 'Population', 'AveOccup', 'Latitude', 'Longitude', 'bedrooms_per_room', 'ocean_proximity_<1H OCEAN', 'ocean_proximity_FAR INLAND', 'ocean_proximity_INLAND', 'ocean_proximity_NEAR BAY']\n",
      "[ -- ] 5.2 income most important: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# assemble the prepared feature names: numeric + the engineered ratio + the one-hot categories\n",
    "ohe = full_pipeline.named_transformers_[\"cat\"]\n",
    "cat_feature_names = list(ohe.get_feature_names_out(cat_columns))\n",
    "prepared_names = num_columns + [\"bedrooms_per_room\"] + cat_feature_names\n",
    "print(f\"{len(prepared_names)} prepared features:\", prepared_names)\n",
    "\n",
    "def top_importances(search, feature_names, k=5):\n",
    "    \"\"\"Return the top-k (name, importance) tuples from the fitted forest, highest first.\"\"\"\n",
    "    importances = search.best_estimator_.named_steps[\"model\"].feature_importances_\n",
    "    assert len(importances) == len(feature_names), (\n",
    "        f\"{len(importances)} importances vs {len(feature_names)} names; \"\n",
    "        \"your prepared_names list is out of sync with the pipeline output\")\n",
    "    # TODO 1: pair each name with its importance, e.g. list(zip(feature_names, importances))\n",
    "    pairs = None\n",
    "    # TODO 2: sort by importance descending and keep the first k\n",
    "    ranked = None\n",
    "    attempted(pairs, ranked)\n",
    "    return [(n, float(v)) for n, v in ranked]\n",
    "\n",
    "def _income_is_top():\n",
    "    top = top_importances(search, prepared_names, k=5)\n",
    "    assert top[0][0] == \"MedInc\", f\"top feature was {top[0][0]!r}, expected 'MedInc'\"\n",
    "    assert all(top[i][1] >= top[i + 1][1] for i in range(len(top) - 1)), \"not sorted descending\"\n",
    "\n",
    "check(\"5.2 income most important\", _income_is_top)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5508f474",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>`zip(names, importances)` pairs them. `sorted(pairs, key=lambda p: p[1], reverse=True)` ranks by the importance (the second element of each pair). Slice `[:k]`.</details>\n",
    "\n",
    "<details><summary>Hint 2 (pseudocode)</summary>\n",
    "\n",
    "```python\n",
    "pairs = list(zip(feature_names, importances))\n",
    "ranked = sorted(pairs, key=lambda p: p[1], reverse=True)[:k]\n",
    "```\n",
    "</details>\n",
    "\n",
    "<details><summary>Help — \"AssertionError: 13 importances vs 14 names\"</summary>Your `prepared_names` has the wrong count. The numeric branch outputs 8 original columns plus 1 ratio = 9, then the one-hot adds one column per category. Re-run the cell that builds `prepared_names`; the order is numeric, then ratio, then categories.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "d3a1e2be",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.049201Z",
     "iopub.status.busy": "2026-06-12T09:33:56.049092Z",
     "iopub.status.idle": "2026-06-12T09:33:56.077994Z",
     "shell.execute_reply": "2026-06-12T09:33:56.077538Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 5.2 income most important\n",
      "  0.353  MedInc\n",
      "  0.121  AveOccup\n",
      "  0.121  Latitude\n",
      "  0.112  bedrooms_per_room\n",
      "  0.103  Longitude\n",
      "  0.065  AveRooms\n",
      "  0.049  HouseAge\n",
      "  0.028  AveBedrms\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines top_importances; the check below re-verifies the reference.\n",
    "def top_importances(search, feature_names, k=5):\n",
    "    importances = search.best_estimator_.named_steps[\"model\"].feature_importances_\n",
    "    pairs = list(zip(feature_names, importances))\n",
    "    ranked = sorted(pairs, key=lambda p: p[1], reverse=True)[:k]\n",
    "    return [(n, float(v)) for n, v in ranked]\n",
    "\n",
    "check(\"5.2 income most important\", _income_is_top, required=True)\n",
    "for name, imp in top_importances(search, prepared_names, k=8):\n",
    "    print(f\"  {imp:.3f}  {name}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2de0659",
   "metadata": {},
   "source": [
    "> **Interpretation.** Income leads by a wide margin, then the geographic coordinates and occupancy. The engineered `bedrooms_per_room` usually places in the middle of the pack, which is the payoff for the work in Exercise 2.2. Three or four features carry most of the signal; the rest are nearly free to drop.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "211ca53b",
   "metadata": {},
   "source": [
    "### 5.3 Error analysis: read the residuals\n",
    "\n",
    "The step most tutorials skip. Predict on the *training* set (we still have not touched test), compute residuals, and look at where the model is wrong. The plot tells you what to fix next.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "8971943f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.078825Z",
     "iopub.status.busy": "2026-06-12T09:33:56.078748Z",
     "iopub.status.idle": "2026-06-12T09:33:56.264895Z",
     "shell.execute_reply": "2026-06-12T09:33:56.264597Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1100x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# train-set residuals (test is still sealed)\n",
    "y_train_pred = search.best_estimator_.predict(X_train)\n",
    "residuals = np.asarray(y_train, dtype=float) - y_train_pred\n",
    "\n",
    "fig, ax = plt.subplots(1, 2, figsize=(11, 4))\n",
    "ax[0].scatter(y_train, residuals, s=5, alpha=0.2, color=\"#1E40FF\")\n",
    "ax[0].axhline(0, color=\"red\", lw=1)\n",
    "ax[0].set_xlabel(\"actual value ($100k)\"); ax[0].set_ylabel(\"residual\"); ax[0].set_title(\"residual vs actual\")\n",
    "ax[1].scatter(X_train[\"MedInc\"], residuals, s=5, alpha=0.2, color=\"#1E40FF\")\n",
    "ax[1].axhline(0, color=\"red\", lw=1)\n",
    "ax[1].set_xlabel(\"median income\"); ax[1].set_ylabel(\"residual\"); ax[1].set_title(\"residual vs income\")\n",
    "plt.tight_layout(); plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "351b317e",
   "metadata": {},
   "source": [
    "> **What is the interpretation of these plots?** <details><summary>Answer</summary>Left: the residuals fan out at the high-value end and there is a hard diagonal wall at the right edge, the fingerprint of the \\$500k cap. The truth was clipped before training, so the model cannot help being wrong there; the honest move is to flag capped districts, not to chase them. Right: residuals are roughly centered across income but with more spread at high income, the same heteroscedasticity. Neither pattern is a coding bug; both are properties of the data the framing should acknowledge.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "be6c9d45",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.266093Z",
     "iopub.status.busy": "2026-06-12T09:33:56.266014Z",
     "iopub.status.idle": "2026-06-12T09:33:56.268892Z",
     "shell.execute_reply": "2026-06-12T09:33:56.268569Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "of the 50 worst-predicted train districts, 62% sit near the $500k cap\n"
     ]
    }
   ],
   "source": [
    "# confirm the worst errors cluster at the target cap, not scattered at random\n",
    "worst = np.argsort(np.abs(residuals))[-50:]\n",
    "frac_at_cap = float((np.asarray(y_train)[worst] >= 4.9).mean())\n",
    "print(f\"of the 50 worst-predicted train districts, {frac_at_cap:.0%} sit near the $500k cap\")\n",
    "assert frac_at_cap > 0.4, \"the cap should dominate the worst errors; if not, inspect the residual plot\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "167827a4",
   "metadata": {},
   "source": [
    "> **Interpretation.** The cap is not a model failure to debug; it is a data property to document. The \"iterate\" arrow in the workflow points back from here: cap-aware features, a censored-regression model, or simply excluding capped districts from the metric are the principled next moves.\n",
    "\n",
    "> **Key takeaways**\n",
    "> - Compare model families on one harness before tuning; the family gap dwarfs the tuning gap.\n",
    "> - Anchor the model on the baseline's own metric — here the \\$30k human-appraiser median AE — and keep the naive mean-guess as a separate sanity floor.\n",
    "> - Feature importances confirm income leads; residuals reveal the cap as the dominant error source.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d1c095e",
   "metadata": {},
   "source": [
    "## Part 6 — Touch the test set once\n",
    "\n",
    "> **Objectives**\n",
    "> - Compute the test metric exactly once, with a bootstrap confidence interval.\n",
    "> - Save the full pipeline (preprocessing + model) as one artifact.\n",
    "> - Reload it and predict on a brand-new district.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3792fcd1",
   "metadata": {},
   "source": [
    "### 6.1 The test set, once, with a confidence interval\n",
    "\n",
    "This is the moment the test set is touched, and the only moment. After this number exists, any further change to the pipeline invalidates it: you would be tuning on the test set, which is leakage by another name. A point estimate alone is not enough. \"Test RMSE 0.47\" means nothing without \"± what?\", so we attach a 95% confidence interval. We use the bootstrap (resample the test errors with replacement, recompute the metric each time), which makes no distributional assumption about the errors.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "422276ff",
   "metadata": {},
   "source": [
    "### Exercise 6.1 — Bootstrap a confidence interval on the test RMSE\n",
    "\n",
    "`Difficulty 3/5 · ~15 min`\n",
    "\n",
    "Fill in `bootstrap_rmse_ci(y_true, y_pred, n_boot, alpha, seed)`. Resample the *indices* with replacement `n_boot` times; for each resample, compute the RMSE on that subset; return the central point estimate and the `(lo, hi)` percentile interval at confidence `1 - alpha`. Pass an explicit seeded RNG so the interval reproduces.\n",
    "\n",
    "> **Stop and think:** why resample indices rather than resampling `(y_true, y_pred)` pairs directly? <details><summary>Answer</summary>They are the same thing if you keep the pairs aligned. Resampling indices guarantees you draw the true value and its matching prediction together. If you resampled the two arrays independently you would shuffle predictions onto the wrong truths and the CI would be garbage.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "39eade6f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.269812Z",
     "iopub.status.busy": "2026-06-12T09:33:56.269713Z",
     "iopub.status.idle": "2026-06-12T09:33:56.274453Z",
     "shell.execute_reply": "2026-06-12T09:33:56.274106Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] 6.1 bootstrap CI: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def bootstrap_rmse_ci(y_true, y_pred, n_boot=1000, alpha=0.05, seed=SEED):\n",
    "    \"\"\"Bootstrap CI for RMSE. Returns (point_rmse, lo, hi) at confidence 1-alpha.\"\"\"\n",
    "    y_true = np.asarray(y_true, dtype=float)\n",
    "    y_pred = np.asarray(y_pred, dtype=float)\n",
    "    n = len(y_true)\n",
    "    boot = np.empty(n_boot)\n",
    "    boot_rng = np.random.default_rng(seed)        # local seeded RNG so the CI reproduces\n",
    "    for b in range(n_boot):\n",
    "        # TODO 1: draw n indices in [0, n) with replacement (boot_rng.integers)\n",
    "        idx = None\n",
    "        attempted(idx)\n",
    "        # TODO 2: RMSE on that resample: sqrt(mean((y_true[idx] - y_pred[idx])**2))\n",
    "        boot[b] = np.sqrt(np.mean((y_true[idx] - y_pred[idx]) ** 2))\n",
    "    point = float(np.sqrt(np.mean((y_true - y_pred) ** 2)))\n",
    "    lo = float(np.percentile(boot, 100 * alpha / 2))\n",
    "    hi = float(np.percentile(boot, 100 * (1 - alpha / 2)))\n",
    "    return point, lo, hi\n",
    "\n",
    "def _ci_sane():\n",
    "    # synthetic: errors are N(0, 0.5), so RMSE ~ 0.5 and the CI must bracket the point estimate\n",
    "    g = np.random.default_rng(123)\n",
    "    yt = g.normal(size=4000)\n",
    "    yp = yt + g.normal(scale=0.5, size=4000)\n",
    "    point, lo, hi = bootstrap_rmse_ci(yt, yp, n_boot=400, seed=SEED)\n",
    "    assert lo < point < hi, f\"point {point:.3f} must lie inside its CI [{lo:.3f}, {hi:.3f}]\"\n",
    "    assert hi - lo < 0.1, f\"CI width {hi - lo:.3f} too wide for n=4000; check you resample n points\"\n",
    "    check_close(point, 0.5, atol=0.05, msg=\"RMSE of N(0,0.5) errors should be near 0.5\")\n",
    "\n",
    "check(\"6.1 bootstrap CI\", _ci_sane)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f84a504",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>One bootstrap resample is `n` integer indices drawn uniformly from `[0, n)` with replacement: `boot_rng.integers(0, n, size=n)`. Index both arrays with the same `idx` so truth and prediction stay paired.</details>\n",
    "\n",
    "<details><summary>Hint 2 (the line)</summary>`idx = boot_rng.integers(0, n, size=n)`. The RMSE line is already written for you.</details>\n",
    "\n",
    "<details><summary>Help — \"the CI is suspiciously wide / the point is outside it\"</summary>If you drew fewer than `n` indices, each resample is smaller than the data and the RMSE varies too much. Draw exactly `n`. If the point estimate falls outside, you computed the point on a resample instead of on the full arrays.</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "5c12316e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.275200Z",
     "iopub.status.busy": "2026-06-12T09:33:56.275120Z",
     "iopub.status.idle": "2026-06-12T09:33:56.287403Z",
     "shell.execute_reply": "2026-06-12T09:33:56.287009Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] 6.1 bootstrap CI\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines bootstrap_rmse_ci; the check below re-verifies the reference.\n",
    "def bootstrap_rmse_ci(y_true, y_pred, n_boot=1000, alpha=0.05, seed=SEED):\n",
    "    y_true = np.asarray(y_true, dtype=float)\n",
    "    y_pred = np.asarray(y_pred, dtype=float)\n",
    "    n = len(y_true)\n",
    "    boot = np.empty(n_boot)\n",
    "    boot_rng = np.random.default_rng(seed)\n",
    "    for b in range(n_boot):\n",
    "        idx = boot_rng.integers(0, n, size=n)\n",
    "        boot[b] = np.sqrt(np.mean((y_true[idx] - y_pred[idx]) ** 2))\n",
    "    point = float(np.sqrt(np.mean((y_true - y_pred) ** 2)))\n",
    "    lo = float(np.percentile(boot, 100 * alpha / 2))\n",
    "    hi = float(np.percentile(boot, 100 * (1 - alpha / 2)))\n",
    "    return point, lo, hi\n",
    "\n",
    "check(\"6.1 bootstrap CI\", _ci_sane, required=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "cc2d5586",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.288203Z",
     "iopub.status.busy": "2026-06-12T09:33:56.288112Z",
     "iopub.status.idle": "2026-06-12T09:33:56.339076Z",
     "shell.execute_reply": "2026-06-12T09:33:56.338726Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test RMSE      = 0.479   95% CI [0.459, 0.500]  (units of $100k)\n",
      "Test MAE       = 0.316\n",
      "Test median AE = 0.203   (robust to the capped districts)  <- THE metric the baseline is on\n",
      "\n",
      "human-appraiser baseline median AE = 0.300  ($30k)\n",
      "model test median AE               = 0.203  ($20,282)\n",
      "-> the model BEATS the $30k appraiser baseline on median AE\n",
      "-> within the $25k tolerance (0.25)\n",
      "\n",
      "(secondary) naive mean-guess RMSE  = 1.151   sanity floor only, NOT the baseline\n"
     ]
    }
   ],
   "source": [
    "# THE TEST SET, TOUCHED ONCE.\n",
    "N_BOOT = 300 if FAST else 1000\n",
    "APPRAISER_MEDIAN_AE = 0.30   # the framing's human-appraiser baseline: $30k median AE, in $100k units\n",
    "TOLERANCE = 0.25             # the product team's tolerance: ~$25k median AE, in $100k units\n",
    "y_test_pred = search.best_estimator_.predict(X_test)        # full pipeline transforms X_test internally\n",
    "abs_err = np.abs(np.asarray(y_test, dtype=float) - y_test_pred)\n",
    "test_rmse, ci_lo, ci_hi = bootstrap_rmse_ci(y_test, y_test_pred, n_boot=N_BOOT)\n",
    "test_median_ae = float(np.median(abs_err))\n",
    "print(f\"Test RMSE      = {test_rmse:.3f}   95% CI [{ci_lo:.3f}, {ci_hi:.3f}]  (units of $100k)\")\n",
    "print(f\"Test MAE       = {abs_err.mean():.3f}\")\n",
    "print(f\"Test median AE = {test_median_ae:.3f}   (robust to the capped districts)  <- THE metric the baseline is on\")\n",
    "print()\n",
    "# Compare to the BASELINE on its own metric: the appraiser baseline is a MEDIAN AE, so compare median AE to it.\n",
    "print(f\"human-appraiser baseline median AE = {APPRAISER_MEDIAN_AE:.3f}  ($30k)\")\n",
    "print(f\"model test median AE               = {test_median_ae:.3f}  (${test_median_ae*100_000:,.0f})\")\n",
    "beats = \"BEATS\" if test_median_ae < APPRAISER_MEDIAN_AE else \"does NOT beat\"\n",
    "print(f\"-> the model {beats} the $30k appraiser baseline on median AE\")\n",
    "print(f\"-> {'within' if test_median_ae < TOLERANCE else 'outside'} the $25k tolerance ({TOLERANCE:.2f})\")\n",
    "print()\n",
    "print(f\"(secondary) naive mean-guess RMSE  = {constant_baseline_rmse(y_train, y_test):.3f}   sanity floor only, NOT the baseline\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9d5743f8",
   "metadata": {},
   "source": [
    "> **Interpretation.** Read the comparison on the right metric. The baseline from the framing is a human-appraiser **median AE of \\$30k** (0.30 in \\$100k units), so the only model number that legitimately compares against it is the model's own **median AE** — about 0.20 here (~\\$20k), which beats the \\$30k appraisers and lands inside the \\$25k tolerance. Do *not* read the model's RMSE (~0.48, ~\\$48k) against the \\$30k baseline: RMSE squares errors and is inflated by the \\$500k-cap tail, so on that number alone the model looks *worse* than the appraisers when on the metric that was actually specified it is clearly better. That gap is the whole lesson — the median AE (~0.20) is far below the MAE (~0.32), which is far below the RMSE (~0.48); the spread between them is the tail of large errors, all living at the cap. The naive mean-guess RMSE (~1.15) is only a sanity floor that proves the model learned something, not the bar that matters. Reporting all three, with the CI, and comparing on the baseline's own metric is the honest summary that a single number hides.\n",
    "\n",
    "> **Caveat:** the CI here is from resampling the test errors; it captures sampling noise in *this* test set. It does not capture distribution shift between 1990 and today, model-selection bias from the choices we made, or the cap. A confidence interval is a statement about the experiment, not a guarantee about the world.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5eda4540",
   "metadata": {},
   "source": [
    "### 6.2 Save the whole pipeline as one artifact\n",
    "\n",
    "A trained sklearn pipeline holds the model *and* every preprocessing statistic. Save them together, load them together, and never re-implement the preprocessing at inference time, because re-implementing it is where the bugs live. We save to a temp file so the notebook leaves no clutter, then reload and predict on a new district to prove the artifact is self-contained.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "eb082f97",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.340196Z",
     "iopub.status.busy": "2026-06-12T09:33:56.340109Z",
     "iopub.status.idle": "2026-06-12T09:33:56.675175Z",
     "shell.execute_reply": "2026-06-12T09:33:56.674821Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "predicted median house value: $238,643\n",
      "save/load round-trip verified; temp file cleaned up\n"
     ]
    }
   ],
   "source": [
    "import joblib, tempfile, os as _os\n",
    "\n",
    "with tempfile.NamedTemporaryFile(suffix=\".joblib\", delete=False) as f:\n",
    "    model_path = f.name\n",
    "joblib.dump(search.best_estimator_, model_path)            # the full Pipeline, preprocessing included\n",
    "loaded = joblib.load(model_path)\n",
    "\n",
    "# a brand-new district: raw columns exactly as the model expects (the pipeline does all the prep)\n",
    "new_district = pd.DataFrame([{\n",
    "    \"MedInc\": 5.0, \"HouseAge\": 25.0, \"AveRooms\": 6.0, \"AveBedrms\": 1.05,\n",
    "    \"Population\": 1200.0, \"AveOccup\": 3.0, \"Latitude\": 34.2, \"Longitude\": -118.5,\n",
    "    \"ocean_proximity\": \"INLAND\",\n",
    "}])\n",
    "predicted = float(loaded.predict(new_district)[0])\n",
    "print(f\"predicted median house value: ${predicted * 100_000:,.0f}\")\n",
    "\n",
    "# the reloaded model must predict identically to the in-memory one (else the artifact is incomplete)\n",
    "assert np.allclose(loaded.predict(new_district), search.best_estimator_.predict(new_district)), \\\n",
    "    \"reloaded model disagrees with the original; the saved artifact is missing state\"\n",
    "_os.remove(model_path)\n",
    "print(\"save/load round-trip verified; temp file cleaned up\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45a71859",
   "metadata": {},
   "source": [
    "> **Interpretation.** One file in, a dollar prediction out, and the reloaded model agrees with the original to floating-point precision. The new district was raw, exactly the shape a production caller would send; the pipeline imputed, engineered, scaled, and encoded it the same way it learned to during training. That is the payoff of putting the firewall in the artifact: inference can never disagree with training, because it *is* training's preprocessing.\n",
    "\n",
    "> **Note:** `joblib.dump` is the sklearn convention (`pickle` works too; `compress=3` for large models). But pickle and joblib execute arbitrary code on load, so never `joblib.load` an untrusted file. The Safety lens returns to this.\n",
    "\n",
    "> **Key takeaways**\n",
    "> - Touch the test set once; a second touch turns evaluation into tuning.\n",
    "> - Report a point estimate with a confidence interval and a robust companion metric.\n",
    "> - Save model and preprocessing as one artifact; the reload must predict identically.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9ce05281",
   "metadata": {},
   "source": [
    "## Safety lens\n",
    "\n",
    "The pipeline you built has failure modes that no textbook demo shows but the day after deployment will. All three are about the gap between the world the model trained on and the world it runs in.\n",
    "\n",
    "**Distribution shift.** This data is from the 1990 census, so it should not be assumed valid for current districts. Log production inputs and compare them with a versioned reference distribution. Statistical tests can be useful diagnostics, but at large sample sizes tiny harmless changes produce small p-values. Alert on sustained, operationally meaningful effect sizes, data-quality failures, or downstream performance changes—not on a p-value alone. The deeper habit: every deployed model carries an implicit \"valid for inputs like X\" clause, so write it down explicitly.\n",
    "\n",
    "**Feature engineering can encode discrimination.** `bedrooms_per_room` looks neutral. But the 1990 geography it derives from reflects decades of housing policy (redlining) that correlates with race. A price model used to inform lending can institutionalize those patterns. The fix is not \"drop the feature\"; it is to audit per-subgroup error rates before shipping and to use reweighing or adversarial debiasing when disparities appear. The non-negotiable habit: never ship a model that affects people without per-subgroup error analysis.\n",
    "\n",
    "**The artifact is an attack surface.** A user-controlled `AveRooms` of zero produces a divide-by-zero in `RatioAdder`; a category never seen in training would be silently dropped (here `handle_unknown=\"ignore\"` makes that explicit, which is the point). And the `joblib` file from 6.2 executes arbitrary code on load: loading an untrusted checkpoint is remote code execution. Two habits: validate inputs at the transformer boundary rather than failing silently, and prefer signed or `safetensors`-style checkpoints in adversarial settings.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "3c89696c",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.676089Z",
     "iopub.status.busy": "2026-06-12T09:33:56.676006Z",
     "iopub.status.idle": "2026-06-12T09:33:56.678810Z",
     "shell.execute_reply": "2026-06-12T09:33:56.678395Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ratio when AveRooms=0: inf  (inf -> downstream NaNs and a confused model)\n",
      "lesson: a transformer should validate inputs, not emit inf/NaN silently\n"
     ]
    }
   ],
   "source": [
    "# a small, honest demo of the divide-by-zero footgun the Safety lens names\n",
    "bad_row = np.array([[0., 0., 0., 1.]])    # AveRooms (col 2) is zero -> 1/0\n",
    "ratio_out = RatioAdder(bedrooms_idx=3, rooms_idx=2).transform(bad_row)\n",
    "print(f\"ratio when AveRooms=0: {ratio_out[0, -1]}  (inf -> downstream NaNs and a confused model)\")\n",
    "assert not np.isfinite(ratio_out[0, -1]), \"AveRooms=0 should produce a non-finite ratio; validate inputs to prevent it\"\n",
    "print(\"lesson: a transformer should validate inputs, not emit inf/NaN silently\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ef266acf",
   "metadata": {},
   "source": [
    "## Test yourself\n",
    "\n",
    "Three parts: concept self-checks with folded answers, auto-checked problems you compute, and a capstone. Every answer is in this notebook; if unsure, re-run that section. Solutions are folded; try before you peek.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4d30243",
   "metadata": {},
   "source": [
    "### Part A — Concepts\n",
    "\n",
    "1. You fit a `StandardScaler` on the full dataset, then split into train and test. What number on your final report is now a lie, and why? <details><summary>Answer</summary>The test RMSE (and any CI on it). The scaler's mean and std absorbed information from the test rows, so the test set is no longer independent of preprocessing. The reported error is optimistically biased; you cannot trust it as an estimate of performance on truly unseen data.</details>\n",
    "\n",
    "2. In Part 4.3 the leaky pipeline reported a clearly positive R² on a target that was pure noise. In one sentence, where exactly did the information leak? <details><summary>Answer</summary>`SelectKBest` was fit on all the data before cross-validation, so it chose the features that happened to correlate with the target in the very rows the CV later treated as held-out, fitting the noise of those folds.</details>\n",
    "\n",
    "3. Why does the random forest beat the single decision tree by so much in the model comparison? <details><summary>Answer</summary>A single tree overfits, low bias but high variance, memorizing its training folds. The forest averages many trees grown on bootstrapped data and random feature subsets, which are decorrelated, so the average cancels much of that variance while keeping the low bias. Bagging trades a little extra compute for a large variance reduction.</details>\n",
    "\n",
    "4. The chapter says tuning bought a smaller gain than model selection. Quote the two RMSE numbers from this notebook that show it. <details><summary>Answer</summary>Linear regression CV RMSE was about 0.72 and the random forest about 0.52: a ~0.24 family gap. Randomized search then moved the forest from ~0.52 to ~0.49, a few hundredths. The family choice was roughly ten times the tuning gain.</details>\n",
    "\n",
    "5. Look at the residual-vs-actual plot from 5.3. What single data property explains the diagonal wall on the right edge, and is it a bug? <details><summary>Answer</summary>The \\$500,001 target cap. House values were clipped at the ceiling before the data was recorded, so the model cannot predict above it and the residuals form a hard diagonal there. It is a data property, not a code bug; the fix is cap-aware handling, not more tuning.</details>\n",
    "\n",
    "6. Why do we report median absolute error alongside RMSE? <details><summary>Answer</summary>RMSE squares errors, so the capped-district tail inflates it disproportionately. Median AE is robust to that tail and reports the typical error. Reporting both shows the typical case and the tail at once; neither alone is honest here.</details>\n",
    "\n",
    "7. What does `handle_unknown=\"ignore\"` on the `OneHotEncoder` do, and why is it a deliberate choice rather than a default to forget about? <details><summary>Answer</summary>A category seen at inference but not at training is encoded as all zeros instead of raising. It keeps inference from crashing on a novel `ocean_proximity` value, but it also means the model treats that district as \"none of the known categories\", which you should monitor for. The choice trades a crash for a silent degradation, which is usually right but must be watched.</details>\n",
    "\n",
    "8. Why is touching the test set twice a leakage problem, even if you never change the model between touches? <details><summary>Answer</summary>The moment a test number influences a decision (keep this feature, try those hyperparameters), the next test number is contaminated by that decision. Repeated touches turn the test set into a validation set you are implicitly optimizing against, and the final number stops estimating unseen performance.</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6e82414",
   "metadata": {},
   "source": [
    "### Part B — Auto-checked problems\n",
    "\n",
    "Three problems. You write the body; the check asserts a property; the folded solution sits after the check. Retrieve the idea, do not copy it.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "902668bf",
   "metadata": {},
   "source": [
    "**Exercise B.1 — RMSE from scratch** · Difficulty 1/5 · ~5 min\n",
    "\n",
    "Implement `rmse(y_true, y_pred)` with no sklearn. It must equal `sklearn.metrics.root_mean_squared_error` on random inputs.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "17b450e4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.679643Z",
     "iopub.status.busy": "2026-06-12T09:33:56.679558Z",
     "iopub.status.idle": "2026-06-12T09:33:56.683589Z",
     "shell.execute_reply": "2026-06-12T09:33:56.683189Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] B.1 rmse: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def rmse(y_true, y_pred):\n",
    "    y_true, y_pred = np.asarray(y_true, dtype=float), np.asarray(y_pred, dtype=float)\n",
    "    # TODO: square the errors, mean them, square-root\n",
    "    result = None\n",
    "    attempted(result)\n",
    "    return float(result)\n",
    "\n",
    "def _rmse_vs_sklearn():\n",
    "    from sklearn.metrics import root_mean_squared_error\n",
    "    g = np.random.default_rng(7)\n",
    "    a, b = g.normal(size=500), g.normal(size=500)\n",
    "    check_close(rmse(a, b), root_mean_squared_error(a, b), msg=\"must match sklearn's RMSE\")\n",
    "\n",
    "check(\"B.1 rmse\", _rmse_vs_sklearn)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1cfc8497",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1</summary>`np.sqrt(np.mean((y_true - y_pred) ** 2))`.</details>\n",
    "<details><summary>Solution</summary>\n",
    "\n",
    "```python\n",
    "def rmse(y_true, y_pred):\n",
    "    y_true, y_pred = np.asarray(y_true, dtype=float), np.asarray(y_pred, dtype=float)\n",
    "    return float(np.sqrt(np.mean((y_true - y_pred) ** 2)))\n",
    "```\n",
    "</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "6cedd5aa",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.684513Z",
     "iopub.status.busy": "2026-06-12T09:33:56.684433Z",
     "iopub.status.idle": "2026-06-12T09:33:56.687457Z",
     "shell.execute_reply": "2026-06-12T09:33:56.687164Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] B.1 rmse\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines rmse; the check re-verifies it.\n",
    "def rmse(y_true, y_pred):\n",
    "    y_true, y_pred = np.asarray(y_true, dtype=float), np.asarray(y_pred, dtype=float)\n",
    "    return float(np.sqrt(np.mean((y_true - y_pred) ** 2)))\n",
    "\n",
    "check(\"B.1 rmse\", _rmse_vs_sklearn, required=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6339e8c3",
   "metadata": {},
   "source": [
    "**Exercise B.2 — Detect leakage by the symptom** · Difficulty 2/5 · ~10 min\n",
    "\n",
    "A clean evaluation has test error close to CV error. Implement `leakage_suspected(cv_rmse, test_rmse, rel_tol=0.25)` that returns `True` when the test RMSE is *more than `rel_tol` better* (lower) than the CV RMSE, the signature of a leak. Equal or worse test error is not suspicious.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "1f1cb38e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.688376Z",
     "iopub.status.busy": "2026-06-12T09:33:56.688308Z",
     "iopub.status.idle": "2026-06-12T09:33:56.691305Z",
     "shell.execute_reply": "2026-06-12T09:33:56.691024Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] B.2 leakage symptom: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def leakage_suspected(cv_rmse, test_rmse, rel_tol=0.25):\n",
    "    \"\"\"True if test_rmse is more than rel_tol fraction LOWER than cv_rmse.\"\"\"\n",
    "    # TODO: test much lower than CV is the alarm. Return cv_rmse - test_rmse > rel_tol * cv_rmse\n",
    "    result = None\n",
    "    attempted(result)\n",
    "    return bool(result)\n",
    "\n",
    "def _leak_logic():\n",
    "    assert leakage_suspected(0.50, 0.30) is True, \"test 0.30 vs CV 0.50 is 40% better -> suspicious\"\n",
    "    assert leakage_suspected(0.50, 0.47) is False, \"a few percent better is normal noise, not a leak\"\n",
    "    assert leakage_suspected(0.50, 0.60) is False, \"test WORSE than CV is overfitting/shift, not leakage\"\n",
    "\n",
    "check(\"B.2 leakage symptom\", _leak_logic)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "609a77d5",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1</summary>The condition is one comparison: how much lower is test than CV, relative to CV? `cv_rmse - test_rmse > rel_tol * cv_rmse`.</details>\n",
    "<details><summary>Solution</summary>\n",
    "\n",
    "```python\n",
    "def leakage_suspected(cv_rmse, test_rmse, rel_tol=0.25):\n",
    "    return bool(cv_rmse - test_rmse > rel_tol * cv_rmse)\n",
    "```\n",
    "</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "5ee4be22",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.692292Z",
     "iopub.status.busy": "2026-06-12T09:33:56.692224Z",
     "iopub.status.idle": "2026-06-12T09:33:56.695224Z",
     "shell.execute_reply": "2026-06-12T09:33:56.694931Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] B.2 leakage symptom\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines leakage_suspected; the check re-verifies it.\n",
    "def leakage_suspected(cv_rmse, test_rmse, rel_tol=0.25):\n",
    "    return bool(cv_rmse - test_rmse > rel_tol * cv_rmse)\n",
    "\n",
    "check(\"B.2 leakage symptom\", _leak_logic, required=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ffca16f9",
   "metadata": {},
   "source": [
    "**Exercise B.3 — Prove the pipeline does not leak** · Difficulty 3/5 · ~12 min\n",
    "\n",
    "The structural claim deserves a structural test. Implement `imputer_median_unchanged_by_test(X_tr, X_te)`: fit a `SimpleImputer(strategy=\"median\")` on `X_tr`, record its learned medians, then call `transform` on `X_te`, and return `True` iff the medians are *unchanged* by the transform. `transform` must never refit. This is the leakage firewall, asserted.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "327b8ac6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.696178Z",
     "iopub.status.busy": "2026-06-12T09:33:56.696098Z",
     "iopub.status.idle": "2026-06-12T09:33:56.700749Z",
     "shell.execute_reply": "2026-06-12T09:33:56.700366Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ -- ] B.3 firewall holds: not attempted yet — fill in the TODO above, then re-run.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def imputer_median_unchanged_by_test(X_tr, X_te):\n",
    "    \"\"\"Fit imputer on train; confirm transforming test does NOT change its learned medians.\"\"\"\n",
    "    imp = SimpleImputer(strategy=\"median\").fit(X_tr)\n",
    "    before = imp.statistics_.copy()\n",
    "    _ = imp.transform(X_te)            # transform the TEST data (output discarded; we watch for side effects)\n",
    "    after = imp.statistics_\n",
    "    # TODO: return whether before and after are identical (np.array_equal of the two median arrays)\n",
    "    unchanged = None\n",
    "    attempted(unchanged)\n",
    "    return bool(unchanged)\n",
    "\n",
    "def _firewall():\n",
    "    g = np.random.default_rng(11)\n",
    "    X_tr = g.normal(size=(200, 3)); X_te = g.normal(loc=5.0, size=(80, 3))  # test has a very different mean\n",
    "    assert imputer_median_unchanged_by_test(X_tr, X_te) is True, (\n",
    "        \"transform() must not refit; the learned medians must survive a transform on shifted test data\")\n",
    "\n",
    "check(\"B.3 firewall holds\", _firewall)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1eca1c1c",
   "metadata": {},
   "source": [
    "<details><summary>Hint 1 (conceptual)</summary>`transform` reuses the statistics from `fit`; it does not recompute them. So `imp.statistics_` after a `transform` equals what it was before. Compare with `np.array_equal(before, after)`.</details>\n",
    "<details><summary>Hint 2 (the line)</summary>`unchanged = np.array_equal(before, after)`. The function returns `bool(unchanged)` for you.</details>\n",
    "<details><summary>Solution</summary>\n",
    "\n",
    "```python\n",
    "def imputer_median_unchanged_by_test(X_tr, X_te):\n",
    "    imp = SimpleImputer(strategy=\"median\").fit(X_tr)\n",
    "    before = imp.statistics_.copy()\n",
    "    _ = imp.transform(X_te)\n",
    "    after = imp.statistics_\n",
    "    return bool(np.array_equal(before, after))\n",
    "```\n",
    "</details>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "3142637a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.701597Z",
     "iopub.status.busy": "2026-06-12T09:33:56.701527Z",
     "iopub.status.idle": "2026-06-12T09:33:56.704424Z",
     "shell.execute_reply": "2026-06-12T09:33:56.704065Z"
    },
    "collapsed": true,
    "jupyter": {
     "source_hidden": true
    },
    "tags": [
     "hide-input"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] B.3 firewall holds\n",
      "the imputer's learned medians survive a transform on shifted test data: the firewall holds\n"
     ]
    }
   ],
   "source": [
    "#@title Solution { display-mode: \"form\" }\n",
    "# solution: redefines imputer_median_unchanged_by_test; the check re-verifies it.\n",
    "def imputer_median_unchanged_by_test(X_tr, X_te):\n",
    "    imp = SimpleImputer(strategy=\"median\").fit(X_tr)\n",
    "    before = imp.statistics_.copy()\n",
    "    _ = imp.transform(X_te)\n",
    "    after = imp.statistics_\n",
    "    return bool(np.array_equal(before, after))\n",
    "\n",
    "check(\"B.3 firewall holds\", _firewall, required=True)\n",
    "print(\"the imputer's learned medians survive a transform on shifted test data: the firewall holds\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85a0e067",
   "metadata": {},
   "source": [
    "### Part C — Capstone: the project, on a dataset you choose\n",
    "\n",
    "Redo this entire chapter on a regression dataset that is not California Housing, end to end. Good choices that load offline: `sklearn.datasets.fetch_openml(\"diamonds\")`, the UCI Bike Sharing set, or `sklearn.datasets.load_diabetes()` (small but complete).\n",
    "\n",
    "**Deliverables**\n",
    "1. A framing cell: objective, metric (with tolerance), and a stated baseline.\n",
    "2. EDA: structure, distributions, and a correlation ranking that names the dominant feature.\n",
    "3. A stratified or otherwise justified split (chronological if it is a time series).\n",
    "4. A `Pipeline` + `ColumnTransformer` with at least one learned step (imputer or scaler) and, if there is a categorical, a one-hot branch.\n",
    "5. A three-family model comparison on one CV harness, then a tuned winner.\n",
    "6. A single test-set evaluation with a bootstrap CI, plus the saved-and-reloaded artifact predicting on one new row.\n",
    "\n",
    "**Self-assessment (pass / partial / fail)**\n",
    "- (a) The framing names objective, metric, and baseline before any model.\n",
    "- (b) No leakage: every learned step is inside the pipeline; the test set is touched exactly once.\n",
    "- (c) At least three model families are compared on the same harness.\n",
    "- (d) The test metric is reported with a confidence interval, not as a bare number.\n",
    "- (e) The saved artifact reloads and predicts identically; the notebook runs top-to-bottom.\n",
    "\n",
    "<details><summary>My solution (reference skeleton, ~30 s on CPU)</summary>\n",
    "\n",
    "```python\n",
    "from sklearn.datasets import load_diabetes\n",
    "from sklearn.model_selection import train_test_split, cross_val_score, RandomizedSearchCV\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from scipy.stats import randint\n",
    "import numpy as np\n",
    "\n",
    "d = load_diabetes(as_frame=True)\n",
    "X, y = d.data, d.target\n",
    "# framing: predict disease progression; metric RMSE; baseline = mean guess\n",
    "Xtr, Xte, ytr, yte = train_test_split(X, y, test_size=0.2, random_state=SEED)\n",
    "\n",
    "# all features are numeric and pre-scaled here, but keep the firewall habit\n",
    "prep = Pipeline([(\"scale\", StandardScaler())])\n",
    "for name, m in [(\"lin\", LinearRegression()),\n",
    "                (\"tree\", DecisionTreeRegressor(random_state=SEED)),\n",
    "                (\"rf\", RandomForestRegressor(n_estimators=80, random_state=SEED))]:\n",
    "    est = Pipeline([(\"prep\", prep), (\"model\", m)])\n",
    "    s = -cross_val_score(est, Xtr, ytr, scoring=\"neg_root_mean_squared_error\", cv=3)\n",
    "    print(f\"{name:5s} CV RMSE {s.mean():.1f}\")\n",
    "\n",
    "rf = Pipeline([(\"prep\", prep), (\"model\", RandomForestRegressor(random_state=SEED))])\n",
    "search = RandomizedSearchCV(rf, {\"model__n_estimators\": randint(40, 120),\n",
    "                                 \"model__max_features\": randint(2, 8)},\n",
    "                            n_iter=8, cv=3, scoring=\"neg_root_mean_squared_error\",\n",
    "                            random_state=SEED).fit(Xtr, ytr)\n",
    "\n",
    "# test once, with the bootstrap CI helper you already wrote above\n",
    "yp = search.best_estimator_.predict(Xte)\n",
    "point, lo, hi = bootstrap_rmse_ci(yte, yp, n_boot=1000)\n",
    "print(f\"test RMSE {point:.1f}  95% CI [{lo:.1f}, {hi:.1f}]\")\n",
    "```\n",
    "\n",
    "The lesson transfers exactly: the family gap dwarfs the tuning gap, and the CI is what makes the test number mean something.</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "18a98f12",
   "metadata": {},
   "source": [
    "## Reflection\n",
    "\n",
    "Write about 150 words on the dumbest bug you hit in this notebook and how you found it. A strong candidate: the leakage demo in 4.3, where a model that predicts pure noise reported a great score, and the only tell was that the number was *too good*. Or the moment a `ColumnTransformer` column count did not match between train and test. Or an off-by-one in the prepared-feature names that made the importances line up with the wrong columns.\n",
    "\n",
    "Name the symptom, the wrong hypothesis you chased first, and the check that actually localized it. Nobody grades this. Writing it is the point: debugging is a skill you build by narrating your own mistakes back to yourself, and an end-to-end project is mostly an exercise in catching the quiet failures before they reach a test number.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc22eb70",
   "metadata": {},
   "source": [
    "## Going further\n",
    "\n",
    "- Géron, *Hands-On ML* 3e, Ch 2 — the full end-to-end housing project this notebook compresses; run it cell by cell next.\n",
    "- scikit-learn user guide, *Common pitfalls and recommended practices* — the canonical write-up of the leakage you watched fabricate a score in 4.3.\n",
    "- scikit-learn, *Column Transformer with Mixed Types* — the pattern behind the firewall in Part 4.\n",
    "- Made-With-ML, *MLOps* (preparation, training, evaluation, monitoring) — the same workflow from a production angle, including the drift monitoring the Safety lens sketched.\n",
    "- Kaggle, *Intro to Machine Learning* — a gentler pass over the same split/fit/evaluate loop, useful as contrast.\n",
    "- fastbook, Ch 2 (*Production*) — an opinionated counter-perspective on deployment and the \"order matters\" notebook-hygiene rules.\n",
    "\n",
    "## What this enables\n",
    "\n",
    "- **Ch 03 — Classification**: the same split/pipeline/evaluate skeleton, with classification metrics replacing RMSE. You will not re-learn the workflow.\n",
    "- **Ch 04 — Training Models**: the linear regression you ran as a baseline becomes a loss you minimize by hand.\n",
    "- **Ch 06 — Ensembles**: the random forest that won here becomes an algorithm you implement and debug.\n",
    "- **Ch 23 — Eval Science**: the bootstrap CI you wrote in 6.1 generalizes into confidence intervals on any metric, plus contamination and judge bias.\n",
    "\n",
    "The gap this notebook leaves: we evaluated on a 1990 snapshot and never watched the world move. The next cell quantifies why that matters, by measuring how far the input distribution would have to shift before a per-feature drift test fires.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "00bb69ca",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-06-12T09:33:56.705079Z",
     "iopub.status.busy": "2026-06-12T09:33:56.704967Z",
     "iopub.status.idle": "2026-06-12T09:33:56.718209Z",
     "shell.execute_reply": "2026-06-12T09:33:56.717802Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ ok ] effect-size drift check: larger sustained shifts trigger review\n",
      "Policy: never page on a p-value alone.\n"
     ]
    }
   ],
   "source": [
    "# scale-up: a tiny taste of the drift monitoring Ch 23/25 make a full chapter\n",
    "ref = np.asarray(X_train[\"MedInc\"], dtype=float)\n",
    "\n",
    "def standardized_mean_shift(reference, current):\n",
    "    \"\"\"Effect size in reference-standard-deviation units.\"\"\"\n",
    "    return abs(float(np.mean(current) - np.mean(reference))) / max(float(np.std(reference)), 1e-12)\n",
    "\n",
    "effects = [standardized_mean_shift(ref, ref + shift) for shift in (0.0, 0.5, 1.5)]\n",
    "assert effects[0] == 0.0 and effects[-1] > effects[1] > 0.2\n",
    "print(\"[ ok ] effect-size drift check: larger sustained shifts trigger review\")\n",
    "print(\"Policy: never page on a p-value alone.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56c92337",
   "metadata": {},
   "source": [
    "---\n",
    "*Built top-to-bottom. If every check above printed `[ ok ]`, you reproduced the chapter end to end.*\n",
    "\n",
    "Total running time: written by CI · Verified on: numpy 2.x, sklearn 1.x, Python 3.12 · 2026-06-11\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "obvix-nb",
   "language": "python",
   "name": "obvix-nb"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  },
  "obvix": {
   "title": "Ch 02 — End-to-End ML Project"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}