jupyter-book/linear_models/module 4 - wrap-up-quizz.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "5bbe0df8",
   "metadata": {},
   "source": [
    "# 🏁 Wrap-up quiz\n",
    "\n",
    "**This quiz requires some programming to be answered.**\n",
    "\n",
    "Open the dataset `house_prices.csv` with the following command:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "de04ff0d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:26:17.964138Z",
     "start_time": "2021-06-03T08:26:17.759258Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "ames_housing = pd.read_csv(\"../datasets/house_prices.csv\", na_values=\"?\")\n",
    "target_name = \"SalePrice\"\n",
    "data = ames_housing.drop(columns=target_name)\n",
    "target = ames_housing[target_name]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dbd2d037",
   "metadata": {},
   "source": [
    "`ames_housing` is a pandas dataframe. The column \"SalePrice\" contains the\n",
    "target variable. Note that we instructed pandas to treat the character \"?\" as a\n",
    "marker for cells with missing values also known as \"null\" values.\n",
    "\n",
    "To simplify this exercise, we will only used the numerical features defined\n",
    "below:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "4e3471fc",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:26:44.898584Z",
     "start_time": "2021-06-03T08:26:44.893663Z"
    }
   },
   "outputs": [],
   "source": [
    "numerical_features = [\n",
    "    \"LotFrontage\", \"LotArea\", \"MasVnrArea\", \"BsmtFinSF1\", \"BsmtFinSF2\",\n",
    "    \"BsmtUnfSF\", \"TotalBsmtSF\", \"1stFlrSF\", \"2ndFlrSF\", \"LowQualFinSF\",\n",
    "    \"GrLivArea\", \"BedroomAbvGr\", \"KitchenAbvGr\", \"TotRmsAbvGrd\", \"Fireplaces\",\n",
    "    \"GarageCars\", \"GarageArea\", \"WoodDeckSF\", \"OpenPorchSF\", \"EnclosedPorch\",\n",
    "    \"3SsnPorch\", \"ScreenPorch\", \"PoolArea\", \"MiscVal\",\n",
    "]\n",
    "\n",
    "data_numerical = data[numerical_features]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45cd1599",
   "metadata": {},
   "source": [
    "Start by fitting a linear regression (`sklearn.linear_model.LinearRegression`).\n",
    "Use a 10-fold cross-validation and pass the argument `return_estimator=True` in\n",
    "`sklearn.model_selection.cross_validate` to access all fitted estimators fitted\n",
    "on each fold. As we saw in the previous notebooks, you will have to use a\n",
    "`sklearn.preprocessing.StandardScaler` to scale the data before passing it to\n",
    "the regressor. Also, some missing data are present in the different columns.\n",
    "You can use a `sklearn.impute.SimpleImputer` with the default parameters to\n",
    "impute missing data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "cc4b4704",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:30:35.771150Z",
     "start_time": "2021-06-03T08:30:35.655177Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.model_selection import cross_validate\n",
    "\n",
    "model = make_pipeline(StandardScaler(), SimpleImputer(), LinearRegression())\n",
    "cv_result = cross_validate(model, data_numerical, target, cv=10, return_estimator=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "7c367250",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:35:50.012029Z",
     "start_time": "2021-06-03T08:35:50.003543Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fold #0 - features selected are: [10 15]\n",
      " and weights are 20942.868431280887\n",
      "Fold #1 - features selected are: [8 6]\n",
      " and weights are 3.820401414396077e+17\n",
      "Fold #2 - features selected are: [10 15]\n",
      " and weights are 19834.47165586813\n",
      "Fold #3 - features selected are: [10 15]\n",
      " and weights are 20243.84416556719\n",
      "Fold #4 - features selected are: [10 15]\n",
      " and weights are 19634.19681674728\n",
      "Fold #5 - features selected are: [10 15]\n",
      " and weights are 22652.887617370136\n",
      "Fold #6 - features selected are: [10 15]\n",
      " and weights are 20200.509719152655\n",
      "Fold #7 - features selected are: [10 15]\n",
      " and weights are 20103.706511184588\n",
      "Fold #8 - features selected are: [ 6 10]\n",
      " and weights are 17152.20750661062\n",
      "Fold #9 - features selected are: [ 3 10]\n",
      " and weights are 1.8041680050201856e+18\n"
     ]
    }
   ],
   "source": [
    "import numpy as np \n",
    "cv_result['estimator']\n",
    "\n",
    "for idx, pipeline in enumerate(cv_result[\"estimator\"]):\n",
    "    print(\n",
    "        f\"Fold #{idx} - features selected are: \"\n",
    "        f\"{np.argsort(pipeline[-1].coef_)[-2:]}\"\n",
    "        f\"\\n and weights are {np.max(abs(pipeline[-1].coef_))}\"\n",
    "    )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "435cc681",
   "metadata": {},
   "source": [
    "# Question 1\n",
    "What is the order of magnitude of the extremum weight values over all the features:\n",
    "\n",
    "- a) 1e4\n",
    "- b) 1e6\n",
    "- c) 1e18\n",
    "\n",
    "_Select a single answer_\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "678a058e",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Repeat the same experiment by fitting a ridge regressor\n",
    "(`sklearn.linear_model.Ridge`) with the default parameter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "49275f42",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:37:00.667080Z",
     "start_time": "2021-06-03T08:37:00.591317Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.linear_model import Ridge\n",
    "from sklearn.model_selection import cross_validate\n",
    "\n",
    "model = make_pipeline(StandardScaler(), SimpleImputer(), Ridge())\n",
    "cv_result = cross_validate(model, data_numerical, target, cv=10, return_estimator=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "1d4809d5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:37:08.634136Z",
     "start_time": "2021-06-03T08:37:08.629913Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fold #0 - features selected are: [10 15]\n",
      " and weights are 20868.215473229386\n",
      "Fold #1 - features selected are: [10 15]\n",
      " and weights are 20330.3048222674\n",
      "Fold #2 - features selected are: [10 15]\n",
      " and weights are 19764.80244161082\n",
      "Fold #3 - features selected are: [10 15]\n",
      " and weights are 20169.142445251662\n",
      "Fold #4 - features selected are: [10 15]\n",
      " and weights are 19570.04868516873\n",
      "Fold #5 - features selected are: [10 15]\n",
      " and weights are 22561.856453251723\n",
      "Fold #6 - features selected are: [10 15]\n",
      " and weights are 20127.805614522495\n",
      "Fold #7 - features selected are: [10 15]\n",
      " and weights are 20032.776933179204\n",
      "Fold #8 - features selected are: [ 6 10]\n",
      " and weights are 17143.069696078583\n",
      "Fold #9 - features selected are: [10 15]\n",
      " and weights are 19143.34269459518\n"
     ]
    }
   ],
   "source": [
    "import numpy as np \n",
    "cv_result['estimator']\n",
    "\n",
    "for idx, pipeline in enumerate(cv_result[\"estimator\"]):\n",
    "    print(\n",
    "        f\"Fold #{idx} - features selected are: \"\n",
    "        f\"{np.argsort(pipeline[-1].coef_)[-2:]}\"\n",
    "        f\"\\n and weights are {np.max(abs(pipeline[-1].coef_))}\"\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "52566fc0",
   "metadata": {},
   "source": [
    "# Question 2\n",
    "What magnitude of the extremum weight values for all features?\n",
    "\n",
    "- a) 1e4\n",
    "- b) 1e6\n",
    "- c) 1e18\n",
    "\n",
    "_Select a single answer_\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1dfd627e",
   "metadata": {},
   "source": [
    "+++"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e62e31fa",
   "metadata": {},
   "source": [
    "# Question 3\n",
    "What are the two most important features used by the ridge regressor? You can\n",
    "make a box-plot of the coefficients across all folds to get a good insight.\n",
    "\n",
    "- a) `\"MiscVal\"` and `\"BsmtFinSF1\"`\n",
    "- b) `\"GarageCars\"` and `\"GrLivArea\"`\n",
    "- c) `\"TotalBsmtSF\"` and `\"GarageCars\"`\n",
    "\n",
    "_Select a single answer_\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "0d6b638d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:44:23.192283Z",
     "start_time": "2021-06-03T08:44:22.920876Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "weights = pd.DataFrame(\n",
    "    [np.abs(est[-1].coef_) for est in cv_result[\"estimator\"]], columns=data[numerical_features].columns)\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "color = {\"whiskers\": \"black\", \"medians\": \"black\", \"caps\": \"black\"}\n",
    "weights.plot.box(color=color, vert=False)\n",
    "_ = plt.title(\"Value of linear regression coefficients\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a01b9f57",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Remove the feature `\"GarageArea\"` from the dataset and repeat the previous\n",
    "experiment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "8fc6bd0f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:53:20.825401Z",
     "start_time": "2021-06-03T08:53:20.458176Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "numerical_features_wo_garage = [\n",
    "    \"LotFrontage\", \"LotArea\", \"MasVnrArea\", \"BsmtFinSF1\", \"BsmtFinSF2\",\n",
    "    \"BsmtUnfSF\", \"TotalBsmtSF\", \"1stFlrSF\", \"2ndFlrSF\", \"LowQualFinSF\",\n",
    "    \"GrLivArea\", \"BedroomAbvGr\", \"KitchenAbvGr\", \"TotRmsAbvGrd\", \"Fireplaces\",\n",
    "    \"GarageCars\", \"WoodDeckSF\", \"OpenPorchSF\", \"EnclosedPorch\",\n",
    "    \"3SsnPorch\", \"ScreenPorch\", \"PoolArea\", \"MiscVal\",\n",
    "]\n",
    "\n",
    "data_wo_garage_numerical = data[numerical_features_wo_garage]\n",
    "\n",
    "\n",
    "\n",
    "cv_result = cross_validate(model, data_wo_garage_numerical, target, cv=10, return_estimator=True)\n",
    "weights = pd.DataFrame(\n",
    "    [np.abs(est[-1].coef_) for est in cv_result[\"estimator\"]], columns=data_wo_garage_numerical.columns)\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "color = {\"whiskers\": \"black\", \"medians\": \"black\", \"caps\": \"black\"}\n",
    "weights.plot.box(color=color, vert=False)\n",
    "_ = plt.title(\"Value of linear regression coefficients\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a67a3a32",
   "metadata": {},
   "source": [
    "# Question 4\n",
    "What is the impact on the weights of removing `\"GarageArea\"` from the dataset?\n",
    "\n",
    "- a) None\n",
    "- b) Change completely the order of the feature importance\n",
    "- c) The variability of the most important feature reduced\n",
    "\n",
    "_Select a single answer_\n",
    "```\n",
    "\n",
    "+++"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "372e59d0",
   "metadata": {},
   "source": [
    "# Question 5\n",
    "What is the reason for observing the previous impact on the most important\n",
    "weight?\n",
    "\n",
    "- a) Both features are correlated and are carrying similar information\n",
    "- b) Removing a feature reduce the noise in the dataset\n",
    "- c) Just some random effects\n",
    "\n",
    "_Select a single answer_\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb5f3fd1",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Now, we will search for the regularization strength that will maximize the\n",
    "statistical performance of our predictive model. Fit a\n",
    "[`sklearn.linear_model.RidgeCV`](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.RidgeCV.html)\n",
    "instead of a `Ridge` regressor pass `alphas=np.logspace(-1, 3, num=30)` to\n",
    "explore the effect of changing the regularization strength."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "e630d4b3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T08:59:46.225227Z",
     "start_time": "2021-06-03T08:59:45.714570Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.linear_model import RidgeCV\n",
    "from sklearn.model_selection import cross_validate\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "model = make_pipeline(StandardScaler(), SimpleImputer(), RidgeCV(alphas=np.logspace(-1, 3, num=30)))\n",
    "cv_result = cross_validate(model, data_numerical, target, cv=10, return_estimator=True)\n",
    "\n",
    "weights = pd.DataFrame(\n",
    "    [np.abs(est[-1].coef_) for est in cv_result[\"estimator\"]], columns=data_numerical.columns)\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "color = {\"whiskers\": \"black\", \"medians\": \"black\", \"caps\": \"black\"}\n",
    "weights.plot.box(color=color, vert=False)\n",
    "_ = plt.title(\"Value of linear regression coefficients\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f0c3db8",
   "metadata": {},
   "source": [
    "# Question 6\n",
    "Are there major differences regarding the most important weights?\n",
    "\n",
    "- a) Yes, the weights order is completely different\n",
    "- b) No, the weights order is very similar\n",
    "\n",
    "_Select a single answer_\n",
    "```\n",
    "\n",
    "+++"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "139009f9",
   "metadata": {},
   "source": [
    "Check the parameter `alpha_` (the regularization strength) for the different\n",
    "ridge regressors obtained on each fold."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "e8056020",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:01:44.195650Z",
     "start_time": "2021-06-03T09:01:44.192158Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 385.6620421163472,\n",
       " 57.361525104486816,\n",
       " 385.6620421163472]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_result\n",
    "[np.abs(est[-1].alpha_) for est in cv_result[\"estimator\"]]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99a5a60a",
   "metadata": {},
   "source": [
    "# Question 7\n",
    "In which range does `alpha_` fall into for most folds?\n",
    "\n",
    "- a) between 0.1 and 1\n",
    "- b) between 1 and 10\n",
    "- c) between 10 and 100\n",
    "- d) between 100 and 1000\n",
    "\n",
    "_Select a single answer_\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "49d1aa17",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Now, we will tackle a classification problem instead of a regression problem.\n",
    "Load the Adult Census dataset with the following snippet of code and we will\n",
    "work only with **numerical features**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "9c909b55",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:02:17.705211Z",
     "start_time": "2021-06-03T09:02:17.641842Z"
    }
   },
   "outputs": [],
   "source": [
    "adult_census = pd.read_csv(\"../datasets/adult-census.csv\")\n",
    "target = adult_census[\"class\"]\n",
    "data = adult_census.select_dtypes([\"integer\", \"floating\"])\n",
    "data = data.drop(columns=[\"education-num\"])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "3a867f92",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:02:39.584072Z",
     "start_time": "2021-06-03T09:02:39.576677Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 48842 entries, 0 to 48841\n",
      "Data columns (total 4 columns):\n",
      " #   Column          Non-Null Count  Dtype\n",
      "---  ------          --------------  -----\n",
      " 0   age             48842 non-null  int64\n",
      " 1   capital-gain    48842 non-null  int64\n",
      " 2   capital-loss    48842 non-null  int64\n",
      " 3   hours-per-week  48842 non-null  int64\n",
      "dtypes: int64(4)\n",
      "memory usage: 1.5 MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7370c2d0",
   "metadata": {},
   "source": [
    "# Question 8\n",
    "How many numerical features are present in the dataset?\n",
    "\n",
    "- a) 3\n",
    "- b) 4\n",
    "- c) 5\n",
    "\n",
    "_Select a single answer_\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "22354768",
   "metadata": {},
   "source": [
    "# Question 9\n",
    "Are there missing values in this dataset?\n",
    "\n",
    "- a) Yes\n",
    "- b) No\n",
    "\n",
    "_Select a single answer_\n",
    "\n",
    "Hint: you can use `df.info()` to get information regarding each column.\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cf3add5b",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Fit a `sklearn.linear_model.LogisticRegression` classifier using a 10-fold\n",
    "cross-validation to assess the performance. Since we are dealing with a linear\n",
    "model, do not forget to scale the data with a `StandardScaler` before training\n",
    "the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "02c3e0a0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:12:18.272361Z",
     "start_time": "2021-06-03T09:12:16.410733Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score is 0.80 +/- 0.00\n"
     ]
    }
   ],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import cross_validate\n",
    "\n",
    "model = make_pipeline(StandardScaler(), LogisticRegression())\n",
    "cv_result = cross_validate(model, data, target, cv=10, return_estimator=True)\n",
    "cv_result\n",
    "\n",
    "print(f'Score is {cv_result[\"test_score\"].mean():0.2f} +/- {cv_result[\"test_score\"].std():0.2f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "600528ea",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:09:22.815987Z",
     "start_time": "2021-06-03T09:09:22.808761Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       " <=50K    37155\n",
       " >50K     11687\n",
       "Name: class, dtype: int64"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "78104d72",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:10:07.149915Z",
     "start_time": "2021-06-03T09:10:07.050889Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of a model predicting only low revenue: 0.761\n"
     ]
    }
   ],
   "source": [
    "from sklearn.dummy import DummyClassifier\n",
    "\n",
    "class_to_predict = \" <=50K\"\n",
    "low_revenue_clf = DummyClassifier(strategy=\"constant\",\n",
    "                                  constant=class_to_predict)\n",
    "low_revenue_clf.fit(data, target)\n",
    "score = low_revenue_clf.score(data, target)\n",
    "print(f\"Accuracy of a model predicting only low revenue: {score:.3f}\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0212804c",
   "metadata": {},
   "source": [
    "# Question 10\n",
    "On average, how much better/worse/similar is the logistic regression  to a dummy\n",
    "classifier that would predict the most frequent class?\n",
    "\n",
    "- a) Worse than a dummy classifier by ~4%\n",
    "- b) Similar to a dummy classifier\n",
    "- c) Better than a dummy classifier by ~4%\n",
    "\n",
    "_Select a single answer_\n",
    "```\n",
    "\n",
    "+++"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "bae4b27f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:16:23.225903Z",
     "start_time": "2021-06-03T09:16:23.131974Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "weights = pd.DataFrame(\n",
    "    np.array([np.abs(est[-1].coef_) for est in cv_result[\"estimator\"]]).reshape(10,-1), columns=data.columns)\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "color = {\"whiskers\": \"black\", \"medians\": \"black\", \"caps\": \"black\"}\n",
    "weights.plot.box(color=color, vert=False)\n",
    "_ = plt.title(\"Value of linear regression coefficients\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "365f3d16",
   "metadata": {},
   "source": [
    "# Question 11\n",
    "What is the most important feature seen by the logistic regression?\n",
    "\n",
    "- a) `\"age\"`\n",
    "- b) `\"capital-gain\"`\n",
    "- c) `\"capital-loss\"`\n",
    "- d) `\"hours-per-week\"`\n",
    "\n",
    "_Select a single answer_\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e2eb4da",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Now, we will work with **both numerical and categorical features**. You can\n",
    "load Adult Census with the following snippet:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "397859b8",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:16:53.333776Z",
     "start_time": "2021-06-03T09:16:53.265077Z"
    }
   },
   "outputs": [],
   "source": [
    "adult_census = pd.read_csv(\"../datasets/adult-census.csv\")\n",
    "target = adult_census[\"class\"]\n",
    "data = adult_census.drop(columns=[\"class\", \"education-num\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "57c598b5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:17:02.443232Z",
     "start_time": "2021-06-03T09:17:02.424611Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 48842 entries, 0 to 48841\n",
      "Data columns (total 12 columns):\n",
      " #   Column          Non-Null Count  Dtype \n",
      "---  ------          --------------  ----- \n",
      " 0   age             48842 non-null  int64 \n",
      " 1   workclass       48842 non-null  object\n",
      " 2   education       48842 non-null  object\n",
      " 3   marital-status  48842 non-null  object\n",
      " 4   occupation      48842 non-null  object\n",
      " 5   relationship    48842 non-null  object\n",
      " 6   race            48842 non-null  object\n",
      " 7   sex             48842 non-null  object\n",
      " 8   capital-gain    48842 non-null  int64 \n",
      " 9   capital-loss    48842 non-null  int64 \n",
      " 10  hours-per-week  48842 non-null  int64 \n",
      " 11  native-country  48842 non-null  object\n",
      "dtypes: int64(4), object(8)\n",
      "memory usage: 4.5+ MB\n"
     ]
    }
   ],
   "source": [
    "data.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f22b16a1",
   "metadata": {},
   "source": [
    "# Question 12\n",
    "Are there missing values in this dataset?\n",
    "\n",
    "- a) Yes\n",
    "- b) No\n",
    "\n",
    "_Select a single answer_\n",
    "\n",
    "Hint: you can use `df.info()` to get information regarding each column.\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc7c42bb",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "Create a predictive model where the categorical data should be one-hot encoded,\n",
    "the numerical data should be scaled, and the predictor used should be a\n",
    "logistic regression classifier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "366db4df",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:20:51.992004Z",
     "start_time": "2021-06-03T09:20:47.274413Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n",
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score is 0.85 +/- 0.00\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/explore/miniconda3/envs/scikit-learn-course/lib/python3.9/site-packages/sklearn/linear_model/_logistic.py:763: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\n",
      "\n",
      "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
      "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  n_iter_i = _check_optimize_result(\n"
     ]
    }
   ],
   "source": [
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import cross_validate\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "\n",
    "from sklearn.compose import make_column_selector as selector\n",
    "\n",
    "categorical_columns_selector = selector(dtype_include=object)\n",
    "categorical_columns = categorical_columns_selector(data)\n",
    "numerical_columns_selector = selector(dtype_exclude=object)\n",
    "numerical_columns = numerical_columns_selector(data)\n",
    "\n",
    "preprocessor = ColumnTransformer(transformers=[\n",
    "    (\"num-preprocessor\", StandardScaler(), numerical_columns),\n",
    "    (\"cat-preprocessor\", OneHotEncoder(handle_unknown='ignore'), categorical_columns)\n",
    "])\n",
    "\n",
    "model = make_pipeline(preprocessor, LogisticRegression())\n",
    "cv_result = cross_validate(model, data, target, cv=10, return_estimator=True)\n",
    "cv_result\n",
    "\n",
    "print(f'Score is {cv_result[\"test_score\"].mean():0.2f} +/- {cv_result[\"test_score\"].std():0.2f}')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b532b365",
   "metadata": {},
   "source": [
    "# Question 13\n",
    "On average, what is the improvement of using the categorical features?\n",
    "\n",
    "- a) It gives similar results\n",
    "- b) It improves the statistical performance by 2.5%\n",
    "- c) it improves the statistical performance by 5%\n",
    "- d) it improves the statistical performance by 7.5%\n",
    "- e) it improves the statistical performance by 10%\n",
    "\n",
    "_Select a single answer_\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f88e40c",
   "metadata": {},
   "source": [
    "+++\n",
    "\n",
    "For the following questions, you can use the following snippet to get the\n",
    "feature names after the preprocessing performed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "8f5feec2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:21:46.180654Z",
     "start_time": "2021-06-03T09:21:46.175915Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ColumnTransformer(transformers=[('num-preprocessor', StandardScaler(),\n",
       "                                 ['age', 'capital-gain', 'capital-loss',\n",
       "                                  'hours-per-week']),\n",
       "                                ('cat-preprocessor',\n",
       "                                 OneHotEncoder(handle_unknown='ignore'),\n",
       "                                 ['workclass', 'education', 'marital-status',\n",
       "                                  'occupation', 'relationship', 'race', 'sex',\n",
       "                                  'native-country'])])"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preprocessor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "2260b399",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:21:59.530758Z",
     "start_time": "2021-06-03T09:21:59.424958Z"
    }
   },
   "outputs": [],
   "source": [
    "preprocessor.fit(data)\n",
    "feature_names = (preprocessor.named_transformers_[\"cat-preprocessor\"]\n",
    "                             .get_feature_names(categorical_columns)).tolist()\n",
    "feature_names += numerical_columns\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f23266cf",
   "metadata": {},
   "source": [
    "There is as many feature names as coefficients in the last step of your\n",
    "predictive pipeline."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "id": "28575090",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-06-03T09:55:49.933972Z",
     "start_time": "2021-06-03T09:55:49.926763Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[27  1]\n",
      "['marital-status_ Married-civ-spouse', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n",
      "1\n",
      "[23  1]\n",
      "['education_ Prof-school', 'workclass_ Federal-gov']\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'education_ Prof-school'"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for cv in cv_result['estimator']:\n",
    "    print(np.argmax(cv[-1].coef_[0]))\n",
    "    print(np.argsort(cv[-1].coef_[0])[-2:])\n",
    "    print([feature_names[i] for i in np.argsort(cv[-1].coef_[0])[-2:]])\n",
    "feature_names[23]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2c7efdf",
   "metadata": {},
   "source": [
    "# Question 14\n",
    "What are the two most important features used by the logistic regressor?\n",
    "\n",
    "- a) `\"hours-per-week\"` and `\"native-country_Columbia\"`\n",
    "- b) `\"workclass_?\"` and `\"naitive_country_?\"`\n",
    "- c) `\"capital-gain\"` and `\"education_Doctorate\"`\n",
    "\n",
    "_Select a single answer_\n",
    "```\n",
    "\n",
    "+++"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f7e73100",
   "metadata": {},
   "source": [
    "# Question 15\n",
    "What is the effect of decreasing the `C` parameter on the coefficients?\n",
    "\n",
    "- a) shrinking the magnitude of the weights towards zeros\n",
    "- b) increasing the magnitude of the weights\n",
    "- c) reducing the weights' variance\n",
    "- d) increasing the weights' variance\n",
    "- e) it has no influence on the weights' variance\n",
    "\n",
    "_Select several answers_\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5101ab8c",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.4"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}