Advanced Learning Algorithms/C2W3_Lab_01_Model_Evaluation_and_Selection.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FvYDy9apGRlt"
   },
   "source": [
    "# Optional Lab: Model Evaluation and Selection\n",
    "\n",
    "Quantifying a learning algorithm's performance and comparing different models are some of the common tasks when applying machine learning to real world applications. In this lab, you will practice doing these using the tips shared in class. Specifically, you will:\n",
    "\n",
    "* split datasets into training, cross validation, and test sets\n",
    "* evaluate regression and classification models\n",
    "* add polynomial features to improve the performance of a linear regression model\n",
    "* compare several neural network architectures\n",
    "\n",
    "This lab will also help you become familiar with the code you'll see in this week's programming assignment. Let's begin!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jj1P1kTmGRlv"
   },
   "source": [
    "## Imports and Lab Setup\n",
    "\n",
    "First, you will import the packages needed for the tasks in this lab. We also included some commands to make the outputs later more readable by reducing verbosity and suppressing non-critical warnings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "xuHD2EOxGRlv"
   },
   "outputs": [],
   "source": [
    "# for array computations and loading data\n",
    "import numpy as np\n",
    "\n",
    "# for building linear regression models and preparing data\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.preprocessing import StandardScaler, PolynomialFeatures\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "# for building and training neural networks\n",
    "import tensorflow as tf\n",
    "\n",
    "# custom functions\n",
    "import utils\n",
    "\n",
    "# reduce display precision on numpy arrays\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# suppress warnings\n",
    "tf.get_logger().setLevel('ERROR')\n",
    "tf.autograph.set_verbosity(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "M_UnQ-tWGRlw"
   },
   "source": [
    "## Regression\n",
    "\n",
    "First, you will be tasked to develop a model for a regression problem. You are given the dataset below consisting of 50 examples of an input feature `x` and its corresponding target `y`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "QLsSum08GRlx",
    "outputId": "0d843000-30bf-4a55-dcd8-cd5e42dc716e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "the shape of the inputs x is: (50, 1)\n",
      "the shape of the targets y is: (50, 1)\n"
     ]
    }
   ],
   "source": [
    "# Load the dataset from the text file\n",
    "data = np.loadtxt('./data/data_w3_ex1.csv', delimiter=',')\n",
    "\n",
    "# Split the inputs and outputs into separate arrays\n",
    "x = data[:,0]\n",
    "y = data[:,1]\n",
    "\n",
    "# Convert 1-D arrays into 2-D because the commands later will require it\n",
    "x = np.expand_dims(x, axis=1)\n",
    "y = np.expand_dims(y, axis=1)\n",
    "\n",
    "print(f\"the shape of the inputs x is: {x.shape}\")\n",
    "print(f\"the shape of the targets y is: {y.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ar2jLTqlpt4x"
   },
   "source": [
    "You can plot the dataset to get an idea of how the target behaves with respect to the input. In case you want to inspect the code, you can find the `plot_dataset()` function in the `utils.py` file outside this notebook."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "VxJqex92GRlx",
    "outputId": "b6d781d1-bff3-4912-abe1-0d7fe5738b08"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the entire dataset\n",
    "utils.plot_dataset(x=x, y=y, title=\"input vs. target\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "WbZdSoJHGRly"
   },
   "source": [
    "## Split the dataset into training, cross validation, and test sets\n",
    "\n",
    "In previous labs, you might have used the entire dataset to train your models. In practice however, it is best to hold out a portion of your data to measure how well your model generalizes to new examples. This will let you know if the model has overfit to your training set.\n",
    "\n",
    "As mentioned in the lecture, it is common to split your data into three parts:\n",
    "\n",
    "* ***training set*** - used to train the model\n",
    "* ***cross validation set (also called validation, development, or dev set)*** - used to evaluate the different model configurations you are choosing from. For example, you can use this to make a decision on what polynomial features to add to your dataset.\n",
    "* ***test set*** - used to give a fair estimate of your chosen model's performance against new examples. This should not be used to make decisions while you are still developing the models.\n",
    "\n",
    "Scikit-learn provides a [`train_test_split`](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html) function to split your data into the parts mentioned above. In the code cell below, you will split the entire dataset into 60% training, 20% cross validation, and 20% test."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "22JKPjmjGRly",
    "outputId": "6f02748d-e8ef-4243-d7ff-a206be7e66f5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "the shape of the training set (input) is: (30, 1)\n",
      "the shape of the training set (target) is: (30, 1)\n",
      "\n",
      "the shape of the cross validation set (input) is: (10, 1)\n",
      "the shape of the cross validation set (target) is: (10, 1)\n",
      "\n",
      "the shape of the test set (input) is: (10, 1)\n",
      "the shape of the test set (target) is: (10, 1)\n"
     ]
    }
   ],
   "source": [
    "# Get 60% of the dataset as the training set. Put the remaining 40% in temporary variables: x_ and y_.\n",
    "x_train, x_, y_train, y_ = train_test_split(x, y, test_size=0.40, random_state=1)\n",
    "\n",
    "# Split the 40% subset above into two: one half for cross validation and the other for the test set\n",
    "x_cv, x_test, y_cv, y_test = train_test_split(x_, y_, test_size=0.50, random_state=1)\n",
    "\n",
    "# Delete temporary variables\n",
    "del x_, y_\n",
    "\n",
    "print(f\"the shape of the training set (input) is: {x_train.shape}\")\n",
    "print(f\"the shape of the training set (target) is: {y_train.shape}\\n\")\n",
    "print(f\"the shape of the cross validation set (input) is: {x_cv.shape}\")\n",
    "print(f\"the shape of the cross validation set (target) is: {y_cv.shape}\\n\")\n",
    "print(f\"the shape of the test set (input) is: {x_test.shape}\")\n",
    "print(f\"the shape of the test set (target) is: {y_test.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "rrgciBNd1HYr"
   },
   "source": [
    "You can plot the dataset again below to see which points were used as training, cross validation, or test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "yGm8JbUUzoUW",
    "outputId": "2f7227ef-9e48-4df5-d913-a5b48e92fe81"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "utils.plot_train_cv_test(x_train, y_train, x_cv, y_cv, x_test, y_test, title=\"input vs. target\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Ayr22PeEGRly"
   },
   "source": [
    "## Fit a linear model\n",
    "\n",
    "Now that you have split the data, one of the first things you can try is to fit a linear model. You will do that in the next sections below."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "z1UdF77SGRlz"
   },
   "source": [
    "### Feature scaling\n",
    "\n",
    "In the previous course of this specialization, you saw that it is usually a good idea to perform feature scaling to help your model converge faster. This is especially true if your input features have widely different ranges of values. Later in this lab, you will be adding polynomial terms so your input features will indeed have different ranges. For example, $x$ runs from around 1600 to 3600, while $x^2$ will run from 2.56 million to 12.96 million. \n",
    "\n",
    "You will only use $x$ for this first model but it's good to practice feature scaling now so you can apply it later. For that, you will use the [`StandardScaler`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html) class from scikit-learn. This computes the z-score of your inputs. As a refresher, the z-score is given by the equation:\n",
    "\n",
    "$$ z = \\frac{x - \\mu}{\\sigma} $$\n",
    "\n",
    "where $\\mu$ is the mean of the feature values and $\\sigma$ is the standard deviation. The code below shows how to prepare the training set using the said class. You can plot the results again to inspect if it still follows the same pattern as before. The new graph should have a reduced range of values for `x`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "QlJ7eQ0TGRlz",
    "outputId": "ed899629-4c5e-468d-d5c8-a4b3349bd62a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computed mean of the training set: 2504.06\n",
      "Computed standard deviation of the training set: 574.85\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Initialize the class\n",
    "scaler_linear = StandardScaler()\n",
    "\n",
    "# Compute the mean and standard deviation of the training set then transform it\n",
    "X_train_scaled = scaler_linear.fit_transform(x_train)\n",
    "\n",
    "print(f\"Computed mean of the training set: {scaler_linear.mean_.squeeze():.2f}\")\n",
    "print(f\"Computed standard deviation of the training set: {scaler_linear.scale_.squeeze():.2f}\")\n",
    "\n",
    "# Plot the results\n",
    "utils.plot_dataset(x=X_train_scaled, y=y_train, title=\"scaled input vs. target\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vzDw7PcJGRlz"
   },
   "source": [
    "### Train the model\n",
    "\n",
    "Next, you will create and train a regression model. For this lab, you will use the [LinearRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html) class but take note that there are other [linear regressors](https://scikit-learn.org/stable/modules/classes.html#classical-linear-regressors) which you can also use."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "DIol3uTkGRlz",
    "outputId": "283e5e1c-9aff-41bb-c2d3-6977861d63bd"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None, normalize=False)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Initialize the class\n",
    "linear_model = LinearRegression()\n",
    "\n",
    "# Train the model\n",
    "linear_model.fit(X_train_scaled, y_train )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "x8LW5leOGRlz"
   },
   "source": [
    "### Evaluate the Model\n",
    "\n",
    "To evaluate the performance of your model, you will measure the error for the training and cross validation sets. For the training error, recall the equation for calculating the mean squared error (MSE):\n",
    "\n",
    "$$J_{train}(\\vec{w}, b) = \\frac{1}{2m_{train}}\\left[\\sum_{i=1}^{m_{train}}(f_{\\vec{w},b}(\\vec{x}_{train}^{(i)}) - y_{train}^{(i)})^2\\right]$$\n",
    "\n",
    "Scikit-learn also has a built-in [`mean_squared_error()`](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.mean_squared_error.html) function that you can use. Take note though that [as per the documentation](https://scikit-learn.org/stable/modules/model_evaluation.html#mean-squared-error), scikit-learn's implementation only divides by `m` and not `2*m`, where `m` is the number of examples. As mentioned in Course 1 of this Specialization (cost function lectures), dividing by `2m` is a convention we will follow but the calculations should still work whether or not you include it. Thus, to match the equation above, you can use the scikit-learn function then divide by 2 as shown below. We also included a for-loop implementation so you can check that it's equal. \n",
    "\n",
    "Another thing to take note: Since you trained the model on scaled values (i.e. using the z-score), you should also feed in the scaled training set instead of its raw values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xmuskcDhfy8z",
    "outputId": "10bb7e65-1723-4ab6-8bbf-f800e8777474"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training MSE (using sklearn function): 406.19374192533155\n",
      "training MSE (for-loop implementation): 406.19374192533155\n"
     ]
    }
   ],
   "source": [
    "# Feed the scaled training set and get the predictions\n",
    "yhat = linear_model.predict(X_train_scaled)\n",
    "\n",
    "# Use scikit-learn's utility function and divide by 2\n",
    "print(f\"training MSE (using sklearn function): {mean_squared_error(y_train, yhat) / 2}\")\n",
    "\n",
    "# for-loop implementation\n",
    "total_squared_error = 0\n",
    "\n",
    "for i in range(len(yhat)):\n",
    "    squared_error_i  = (yhat[i] - y_train[i])**2\n",
    "    total_squared_error += squared_error_i                                              \n",
    "\n",
    "mse = total_squared_error / (2*len(yhat))\n",
    "\n",
    "print(f\"training MSE (for-loop implementation): {mse.squeeze()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "lujrQns9f-vu"
   },
   "source": [
    "You can then compute the MSE for the cross validation set with basically the same equation:\n",
    "\n",
    "$$J_{cv}(\\vec{w}, b) = \\frac{1}{2m_{cv}}\\left[\\sum_{i=1}^{m_{cv}}(f_{\\vec{w},b}(\\vec{x}_{cv}^{(i)}) - y_{cv}^{(i)})^2\\right]$$\n",
    "\n",
    "As with the training set, you will also want to scale the cross validation set. An *important* thing to note when using the z-score is you have to use the mean and standard deviation of the **training set** when scaling the cross validation set. This is to ensure that your input features are transformed as expected by the model. One way to gain intuition is with this scenario:\n",
    "\n",
    "* Say that your training set has an input feature equal to `500` which is scaled down to `0.5` using the z-score.\n",
    "* After training, your model is able to accurately map this scaled input `x=0.5` to the target output `y=300`. \n",
    "* Now let's say that you deployed this model and one of your users fed it a sample equal to `500`. \n",
    "* If you get this input sample's z-score using any other values of the mean and standard deviation, then it might not be scaled to `0.5` and your model will most likely make a wrong prediction (i.e. not equal to `y=300`). \n",
    "\n",
    "You will scale the cross validation set below by using the same `StandardScaler` you used earlier but only calling its [`transform()`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html#sklearn.preprocessing.StandardScaler.transform) method instead of [`fit_transform()`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html#sklearn.preprocessing.StandardScaler.fit_transform)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "JJQbmq70GRl0",
    "outputId": "7b75d664-7cf0-4a27-c79c-3236857f444b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean used to scale the CV set: 2504.06\n",
      "Standard deviation used to scale the CV set: 574.85\n",
      "Cross validation MSE: 551.7789026952216\n"
     ]
    }
   ],
   "source": [
    "# Scale the cross validation set using the mean and standard deviation of the training set\n",
    "X_cv_scaled = scaler_linear.transform(x_cv)\n",
    "\n",
    "print(f\"Mean used to scale the CV set: {scaler_linear.mean_.squeeze():.2f}\")\n",
    "print(f\"Standard deviation used to scale the CV set: {scaler_linear.scale_.squeeze():.2f}\")\n",
    "\n",
    "# Feed the scaled cross validation set\n",
    "yhat = linear_model.predict(X_cv_scaled)\n",
    "\n",
    "# Use scikit-learn's utility function and divide by 2\n",
    "print(f\"Cross validation MSE: {mean_squared_error(y_cv, yhat) / 2}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BZtb_blcGRl0"
   },
   "source": [
    "## Adding Polynomial Features\n",
    "\n",
    "From the graphs earlier, you may have noticed that the target `y` rises more sharply at smaller values of `x` compared to higher ones. A straight line might not be the best choice because the target `y` seems to flatten out as `x` increases. Now that you have these values of the training and cross validation MSE from the linear model, you can try adding polynomial features to see if you can get a better performance. The code will mostly be the same but with a few extra preprocessing steps. Let's see that below."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create the additional features\n",
    "\n",
    "First, you will generate the polynomial features from your training set. The code below demonstrates how to do this using the [`PolynomialFeatures`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.PolynomialFeatures.html) class. It will create a new input feature which has the squared values of the input `x` (i.e. degree=2)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "rt1FCgQOGRl0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[3.32e+03 1.11e+07]\n",
      " [2.34e+03 5.50e+06]\n",
      " [3.49e+03 1.22e+07]\n",
      " [2.63e+03 6.92e+06]\n",
      " [2.59e+03 6.71e+06]]\n"
     ]
    }
   ],
   "source": [
    "# Instantiate the class to make polynomial features\n",
    "poly = PolynomialFeatures(degree=2, include_bias=False)\n",
    "\n",
    "# Compute the number of features and transform the training set\n",
    "X_train_mapped = poly.fit_transform(x_train)\n",
    "\n",
    "# Preview the first 5 elements of the new training set. Left column is `x` and right column is `x^2`\n",
    "# Note: The `e+<number>` in the output denotes how many places the decimal point should \n",
    "# be moved. For example, `3.24e+03` is equal to `3240`\n",
    "print(X_train_mapped[:5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You will then scale the inputs as before to narrow down the range of values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "dVJUmPQKGRl1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 1.43  1.47]\n",
      " [-0.28 -0.36]\n",
      " [ 1.71  1.84]\n",
      " [ 0.22  0.11]\n",
      " [ 0.15  0.04]]\n"
     ]
    }
   ],
   "source": [
    "# Instantiate the class\n",
    "scaler_poly = StandardScaler()\n",
    "\n",
    "# Compute the mean and standard deviation of the training set then transform it\n",
    "X_train_mapped_scaled = scaler_poly.fit_transform(X_train_mapped)\n",
    "\n",
    "# Preview the first 5 elements of the scaled training set.\n",
    "print(X_train_mapped_scaled[:5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can then proceed to train the model. After that, you will measure the model's performance against the cross validation set. Like before, you should make sure to perform the same transformations as you did in the training set. You will add the same number of polynomial features then scale the range of values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "WO7rK-f9GRl1",
    "outputId": "12a61725-7eab-40c0-91c2-2acc9f1c5d2a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training MSE: 49.11160933402521\n",
      "Cross validation MSE: 87.69841211111924\n"
     ]
    }
   ],
   "source": [
    "# Initialize the class\n",
    "model = LinearRegression()\n",
    "\n",
    "# Train the model\n",
    "model.fit(X_train_mapped_scaled, y_train )\n",
    "\n",
    "# Compute the training MSE\n",
    "yhat = model.predict(X_train_mapped_scaled)\n",
    "print(f\"Training MSE: {mean_squared_error(y_train, yhat) / 2}\")\n",
    "\n",
    "# Add the polynomial features to the cross validation set\n",
    "X_cv_mapped = poly.transform(x_cv)\n",
    "\n",
    "# Scale the cross validation set using the mean and standard deviation of the training set\n",
    "X_cv_mapped_scaled = scaler_poly.transform(X_cv_mapped)\n",
    "\n",
    "# Compute the cross validation MSE\n",
    "yhat = model.predict(X_cv_mapped_scaled)\n",
    "print(f\"Cross validation MSE: {mean_squared_error(y_cv, yhat) / 2}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You'll notice that the MSEs are significantly better for both the training and cross validation set when you added the 2nd order polynomial. You may want to introduce more polynomial terms and see which one gives the best performance. As shown in class, you can have 10 different models like this:\n",
    "\n",
    "<img src='images/C2_W3_poly.png' width=50%>\n",
    "\n",
    "You can create a loop that contains all the steps in the previous code cells. Here is one implementation that adds polynomial features up to degree=10. We'll plot it at the end to make it easier to compare the results for each model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Initialize lists containing the lists, models, and scalers\n",
    "train_mses = []\n",
    "cv_mses = []\n",
    "models = []\n",
    "scalers = []\n",
    "\n",
    "# Loop over 10 times. Each adding one more degree of polynomial higher than the last.\n",
    "for degree in range(1,11):\n",
    "    \n",
    "    # Add polynomial features to the training set\n",
    "    poly = PolynomialFeatures(degree, include_bias=False)\n",
    "    X_train_mapped = poly.fit_transform(x_train)\n",
    "    \n",
    "    # Scale the training set\n",
    "    scaler_poly = StandardScaler()\n",
    "    X_train_mapped_scaled = scaler_poly.fit_transform(X_train_mapped)\n",
    "    scalers.append(scaler_poly)\n",
    "    \n",
    "    # Create and train the model\n",
    "    model = LinearRegression()\n",
    "    model.fit(X_train_mapped_scaled, y_train )\n",
    "    models.append(model)\n",
    "    \n",
    "    # Compute the training MSE\n",
    "    yhat = model.predict(X_train_mapped_scaled)\n",
    "    train_mse = mean_squared_error(y_train, yhat) / 2\n",
    "    train_mses.append(train_mse)\n",
    "    \n",
    "    # Add polynomial features and scale the cross validation set\n",
    "    poly = PolynomialFeatures(degree, include_bias=False)\n",
    "    X_cv_mapped = poly.fit_transform(x_cv)\n",
    "    X_cv_mapped_scaled = scaler_poly.transform(X_cv_mapped)\n",
    "    \n",
    "    # Compute the cross validation MSE\n",
    "    yhat = model.predict(X_cv_mapped_scaled)\n",
    "    cv_mse = mean_squared_error(y_cv, yhat) / 2\n",
    "    cv_mses.append(cv_mse)\n",
    "    \n",
    "# Plot the results\n",
    "degrees=range(1,11)\n",
    "utils.plot_train_cv_mses(degrees, train_mses, cv_mses, title=\"degree of polynomial vs. train and CV MSEs\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "R83nSGpXGRl2"
   },
   "source": [
    "### Choosing the best model\n",
    "\n",
    "When selecting a model, you want to choose one that performs well both on the training and cross validation set. It implies that it is able to learn the patterns from your training set without overfitting. If you used the defaults in this lab, you will notice a sharp drop in cross validation error from the models with degree=1 to degree=2. This is followed by a relatively flat line up to degree=5. After that, however, the cross validation error is generally getting worse as you add more polynomial features. Given these, you can decide to use the model with the lowest `cv_mse` as the one best suited for your application."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Lowest CV MSE is found in the model with degree=4\n"
     ]
    }
   ],
   "source": [
    "# Get the model with the lowest CV MSE (add 1 because list indices start at 0)\n",
    "# This also corresponds to the degree of the polynomial added\n",
    "degree = np.argmin(cv_mses) + 1\n",
    "print(f\"Lowest CV MSE is found in the model with degree={degree}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can then publish the generalization error by computing the test set's MSE. As usual, you should transform this data the same way you did with the training and cross validation sets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training MSE: 47.15\n",
      "Cross Validation MSE: 79.43\n",
      "Test MSE: 104.63\n"
     ]
    }
   ],
   "source": [
    "# Add polynomial features to the test set\n",
    "poly = PolynomialFeatures(degree, include_bias=False)\n",
    "X_test_mapped = poly.fit_transform(x_test)\n",
    "\n",
    "# Scale the test set\n",
    "X_test_mapped_scaled = scalers[degree-1].transform(X_test_mapped)\n",
    "\n",
    "# Compute the test MSE\n",
    "yhat = models[degree-1].predict(X_test_mapped_scaled)\n",
    "test_mse = mean_squared_error(y_test, yhat) / 2\n",
    "\n",
    "print(f\"Training MSE: {train_mses[degree-1]:.2f}\")\n",
    "print(f\"Cross Validation MSE: {cv_mses[degree-1]:.2f}\")\n",
    "print(f\"Test MSE: {test_mse:.2f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "m5uStJF3GRl2"
   },
   "source": [
    "## Neural Networks\n",
    "\n",
    "The same model selection process can also be used when choosing between different neural network architectures. In this section, you will create the models shown below and apply it to the same regression task above.\n",
    "\n",
    "<img src='images/C2_W3_NN_Arch.png' width=40%>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prepare the Data\n",
    "\n",
    "You will use the same training, cross validation, and test sets you generated in the previous section. From earlier lectures in this course, you may have known that neural networks can learn non-linear relationships so you can opt to skip adding polynomial features. The code is still included below in case you want to try later and see what effect it will have on your results. The default `degree` is set to `1` to indicate that it will just use `x_train`, `x_cv`, and `x_test` as is (i.e. without any additional polynomial features)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "-armALWpGRl2"
   },
   "outputs": [],
   "source": [
    "# Add polynomial features\n",
    "degree = 1\n",
    "poly = PolynomialFeatures(degree, include_bias=False)\n",
    "X_train_mapped = poly.fit_transform(x_train)\n",
    "X_cv_mapped = poly.transform(x_cv)\n",
    "X_test_mapped = poly.transform(x_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, you will scale the input features to help gradient descent converge faster. Again, notice that you are using the mean and standard deviation computed from the training set by just using `transform()` in the cross validation and test sets instead of `fit_transform()`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "hBn-0m0IGRl2"
   },
   "outputs": [],
   "source": [
    "# Scale the features using the z-score\n",
    "scaler = StandardScaler()\n",
    "X_train_mapped_scaled = scaler.fit_transform(X_train_mapped)\n",
    "X_cv_mapped_scaled = scaler.transform(X_cv_mapped)\n",
    "X_test_mapped_scaled = scaler.transform(X_test_mapped)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build and train the models\n",
    "\n",
    "You will then create the neural network architectures shown earlier. The code is provided in the `build_models()` function in the `utils.py` file in case you want to inspect or modify it. You will use that in the loop below then proceed to train the models. For each model, you will also record the training and cross validation errors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model_1...\n",
      "Done!\n",
      "\n",
      "Training model_2...\n",
      "Done!\n",
      "\n",
      "Training model_3...\n",
      "Done!\n",
      "\n",
      "RESULTS:\n",
      "Model 1: Training MSE: 73.44, CV MSE: 113.87\n",
      "Model 2: Training MSE: 73.40, CV MSE: 112.28\n",
      "Model 3: Training MSE: 44.56, CV MSE: 88.51\n"
     ]
    }
   ],
   "source": [
    "# Initialize lists that will contain the errors for each model\n",
    "nn_train_mses = []\n",
    "nn_cv_mses = []\n",
    "\n",
    "# Build the models\n",
    "nn_models = utils.build_models()\n",
    "\n",
    "# Loop over the the models\n",
    "for model in nn_models:\n",
    "    \n",
    "    # Setup the loss and optimizer\n",
    "    model.compile(\n",
    "    loss='mse',\n",
    "    optimizer=tf.keras.optimizers.Adam(learning_rate=0.1),\n",
    "    )\n",
    "\n",
    "    print(f\"Training {model.name}...\")\n",
    "    \n",
    "    # Train the model\n",
    "    model.fit(\n",
    "        X_train_mapped_scaled, y_train,\n",
    "        epochs=300,\n",
    "        verbose=0\n",
    "    )\n",
    "    \n",
    "    print(\"Done!\\n\")\n",
    "\n",
    "    \n",
    "    # Record the training MSEs\n",
    "    yhat = model.predict(X_train_mapped_scaled)\n",
    "    train_mse = mean_squared_error(y_train, yhat) / 2\n",
    "    nn_train_mses.append(train_mse)\n",
    "    \n",
    "    # Record the cross validation MSEs \n",
    "    yhat = model.predict(X_cv_mapped_scaled)\n",
    "    cv_mse = mean_squared_error(y_cv, yhat) / 2\n",
    "    nn_cv_mses.append(cv_mse)\n",
    "\n",
    "    \n",
    "# print results\n",
    "print(\"RESULTS:\")\n",
    "for model_num in range(len(nn_train_mses)):\n",
    "    print(\n",
    "        f\"Model {model_num+1}: Training MSE: {nn_train_mses[model_num]:.2f}, \" +\n",
    "        f\"CV MSE: {nn_cv_mses[model_num]:.2f}\"\n",
    "        )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the recorded errors, you can decide which is the best model for your application. Look at the results above and see if you agree with the selected `model_num` below. Finally, you will compute the test error to estimate how well it generalizes to new examples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Selected Model: 3\n",
      "Training MSE: 44.56\n",
      "Cross Validation MSE: 88.51\n",
      "Test MSE: 87.77\n"
     ]
    }
   ],
   "source": [
    "# Select the model with the lowest CV MSE\n",
    "model_num = 3\n",
    "\n",
    "# Compute the test MSE\n",
    "yhat = nn_models[model_num-1].predict(X_test_mapped_scaled)\n",
    "test_mse = mean_squared_error(y_test, yhat) / 2\n",
    "\n",
    "print(f\"Selected Model: {model_num}\")\n",
    "print(f\"Training MSE: {nn_train_mses[model_num-1]:.2f}\")\n",
    "print(f\"Cross Validation MSE: {nn_cv_mses[model_num-1]:.2f}\")\n",
    "print(f\"Test MSE: {test_mse:.2f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classification\n",
    "\n",
    "In this last part of the lab, you will practice model evaluation and selection on a classification task. The process will be similar, with the main difference being the computation of the errors. You will see that in the following sections."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load the Dataset\n",
    "\n",
    "First, you will load a dataset for a binary classification task. It has 200 examples of two input features (`x1` and `x2`), and a target `y` of either `0` or `1`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "the shape of the inputs x is: (200, 2)\n",
      "the shape of the targets y is: (200, 1)\n"
     ]
    }
   ],
   "source": [
    "# Load the dataset from a text file\n",
    "data = np.loadtxt('./data/data_w3_ex2.csv', delimiter=',')\n",
    "\n",
    "# Split the inputs and outputs into separate arrays\n",
    "x_bc = data[:,:-1]\n",
    "y_bc = data[:,-1]\n",
    "\n",
    "# Convert y into 2-D because the commands later will require it (x is already 2-D)\n",
    "y_bc = np.expand_dims(y_bc, axis=1)\n",
    "\n",
    "print(f\"the shape of the inputs x is: {x_bc.shape}\")\n",
    "print(f\"the shape of the targets y is: {y_bc.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can plot the dataset to examine how the examples are separated."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "utils.plot_bc_dataset(x=x_bc, y=y_bc, title=\"x1 vs. x2\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split and prepare the dataset\n",
    "\n",
    "Next, you will generate the training, cross validation, and test sets. You will use the same 60/20/20 proportions as before. You will also scale the features as you did in the previous section."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "the shape of the training set (input) is: (120, 2)\n",
      "the shape of the training set (target) is: (120, 1)\n",
      "\n",
      "the shape of the cross validation set (input) is: (40, 2)\n",
      "the shape of the cross validation set (target) is: (40, 1)\n",
      "\n",
      "the shape of the test set (input) is: (40, 2)\n",
      "the shape of the test set (target) is: (40, 1)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# Get 60% of the dataset as the training set. Put the remaining 40% in temporary variables.\n",
    "x_bc_train, x_, y_bc_train, y_ = train_test_split(x_bc, y_bc, test_size=0.40, random_state=1)\n",
    "\n",
    "# Split the 40% subset above into two: one half for cross validation and the other for the test set\n",
    "x_bc_cv, x_bc_test, y_bc_cv, y_bc_test = train_test_split(x_, y_, test_size=0.50, random_state=1)\n",
    "\n",
    "# Delete temporary variables\n",
    "del x_, y_\n",
    "\n",
    "print(f\"the shape of the training set (input) is: {x_bc_train.shape}\")\n",
    "print(f\"the shape of the training set (target) is: {y_bc_train.shape}\\n\")\n",
    "print(f\"the shape of the cross validation set (input) is: {x_bc_cv.shape}\")\n",
    "print(f\"the shape of the cross validation set (target) is: {y_bc_cv.shape}\\n\")\n",
    "print(f\"the shape of the test set (input) is: {x_bc_test.shape}\")\n",
    "print(f\"the shape of the test set (target) is: {y_bc_test.shape}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Scale the features\n",
    "\n",
    "# Initialize the class\n",
    "scaler_linear = StandardScaler()\n",
    "\n",
    "# Compute the mean and standard deviation of the training set then transform it\n",
    "x_bc_train_scaled = scaler_linear.fit_transform(x_bc_train)\n",
    "x_bc_cv_scaled = scaler_linear.transform(x_bc_cv)\n",
    "x_bc_test_scaled = scaler_linear.transform(x_bc_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluating the error for classification models\n",
    "\n",
    "In the previous sections on regression models, you used the mean squared error to measure how well your model is doing. For classification, you can get a similar metric by getting the fraction of the data that the model has misclassified. For example, if your model made wrong predictions for 2 samples out of 5, then you will report an error of `40%` or `0.4`. The code below demonstrates this using a for-loop and also with Numpy's [`mean()`](https://numpy.org/doc/stable/reference/generated/numpy.mean.html) function. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "probabilities: [0.2 0.6 0.7 0.3 0.8]\n",
      "predictions with threshold=0.5: [0 1 1 0 1]\n",
      "targets: [1 1 1 1 1]\n",
      "fraction of misclassified data (for-loop): 0.4\n",
      "fraction of misclassified data (with np.mean()): 0.4\n"
     ]
    }
   ],
   "source": [
    "# Sample model output\n",
    "probabilities = np.array([0.2, 0.6, 0.7, 0.3, 0.8])\n",
    "\n",
    "# Apply a threshold to the model output. If greater than 0.5, set to 1. Else 0.\n",
    "predictions = np.where(probabilities >= 0.5, 1, 0)\n",
    "\n",
    "# Ground truth labels\n",
    "ground_truth = np.array([1, 1, 1, 1, 1])\n",
    "\n",
    "# Initialize counter for misclassified data\n",
    "misclassified = 0\n",
    "\n",
    "# Get number of predictions\n",
    "num_predictions = len(predictions)\n",
    "\n",
    "# Loop over each prediction\n",
    "for i in range(num_predictions):\n",
    "    \n",
    "    # Check if it matches the ground truth\n",
    "    if predictions[i] != ground_truth[i]:\n",
    "        \n",
    "        # Add one to the counter if the prediction is wrong\n",
    "        misclassified += 1\n",
    "\n",
    "# Compute the fraction of the data that the model misclassified\n",
    "fraction_error = misclassified/num_predictions\n",
    "\n",
    "print(f\"probabilities: {probabilities}\")\n",
    "print(f\"predictions with threshold=0.5: {predictions}\")\n",
    "print(f\"targets: {ground_truth}\")\n",
    "print(f\"fraction of misclassified data (for-loop): {fraction_error}\")\n",
    "print(f\"fraction of misclassified data (with np.mean()): {np.mean(predictions != ground_truth)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build and train the model\n",
    "\n",
    "You will use the same neural network architectures in the previous section so you can call the `build_models()` function again to create new instances of these models. \n",
    "\n",
    "You will follow the recommended approach mentioned last week where you use a `linear` activation for the output layer (instead of `sigmoid`) then set `from_logits=True` when declaring the loss function of the model. You will use the [binary crossentropy loss](https://www.tensorflow.org/api_docs/python/tf/keras/losses/BinaryCrossentropy) because this is a binary classification problem.\n",
    "\n",
    "After training, you will use a [sigmoid function](https://www.tensorflow.org/api_docs/python/tf/math/sigmoid) to convert the model outputs into probabilities. From there, you can set a threshold and get the fraction of misclassified examples from the training and cross validation sets.\n",
    "\n",
    "You can see all these in the code cell below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model_1...\n",
      "Done!\n",
      "\n",
      "Training model_2...\n",
      "Done!\n",
      "\n",
      "Training model_3...\n",
      "Done!\n",
      "\n",
      "Model 1: Training Set Classification Error: 0.05833, CV Set Classification Error: 0.17500\n",
      "Model 2: Training Set Classification Error: 0.06667, CV Set Classification Error: 0.15000\n",
      "Model 3: Training Set Classification Error: 0.05000, CV Set Classification Error: 0.15000\n"
     ]
    }
   ],
   "source": [
    "# Initialize lists that will contain the errors for each model\n",
    "nn_train_error = []\n",
    "nn_cv_error = []\n",
    "\n",
    "# Build the models\n",
    "models_bc = utils.build_models()\n",
    "\n",
    "# Loop over each model\n",
    "for model in models_bc:\n",
    "    \n",
    "    # Setup the loss and optimizer\n",
    "    model.compile(\n",
    "    loss=tf.keras.losses.BinaryCrossentropy(from_logits=True),\n",
    "    optimizer=tf.keras.optimizers.Adam(learning_rate=0.01),\n",
    "    )\n",
    "\n",
    "    print(f\"Training {model.name}...\")\n",
    "\n",
    "    # Train the model\n",
    "    model.fit(\n",
    "        x_bc_train_scaled, y_bc_train,\n",
    "        epochs=200,\n",
    "        verbose=0\n",
    "    )\n",
    "    \n",
    "    print(\"Done!\\n\")\n",
    "    \n",
    "    # Set the threshold for classification\n",
    "    threshold = 0.5\n",
    "    \n",
    "    # Record the fraction of misclassified examples for the training set\n",
    "    yhat = model.predict(x_bc_train_scaled)\n",
    "    yhat = tf.math.sigmoid(yhat)\n",
    "    yhat = np.where(yhat >= threshold, 1, 0)\n",
    "    train_error = np.mean(yhat != y_bc_train)\n",
    "    nn_train_error.append(train_error)\n",
    "\n",
    "    # Record the fraction of misclassified examples for the cross validation set\n",
    "    yhat = model.predict(x_bc_cv_scaled)\n",
    "    yhat = tf.math.sigmoid(yhat)\n",
    "    yhat = np.where(yhat >= threshold, 1, 0)\n",
    "    cv_error = np.mean(yhat != y_bc_cv)\n",
    "    nn_cv_error.append(cv_error)\n",
    "\n",
    "# Print the result\n",
    "for model_num in range(len(nn_train_error)):\n",
    "    print(\n",
    "        f\"Model {model_num+1}: Training Set Classification Error: {nn_train_error[model_num]:.5f}, \" +\n",
    "        f\"CV Set Classification Error: {nn_cv_error[model_num]:.5f}\"\n",
    "        )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the output above, you can choose which one performed best. If there is a tie on the cross validation set error, then you can pick the one with the lower training set error. Finally, you can compute the test error to report the model's generalization error."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Selected Model: 3\n",
      "Training Set Classification Error: 0.0500\n",
      "CV Set Classification Error: 0.1500\n",
      "Test Set Classification Error: 0.1750\n"
     ]
    }
   ],
   "source": [
    "# Select the model with the lowest error\n",
    "model_num = 3\n",
    "\n",
    "# Compute the test error\n",
    "yhat = models_bc[model_num-1].predict(x_bc_test_scaled)\n",
    "yhat = tf.math.sigmoid(yhat)\n",
    "yhat = np.where(yhat >= threshold, 1, 0)\n",
    "nn_test_error = np.mean(yhat != y_bc_test)\n",
    "\n",
    "print(f\"Selected Model: {model_num}\")\n",
    "print(f\"Training Set Classification Error: {nn_train_error[model_num-1]:.4f}\")\n",
    "print(f\"CV Set Classification Error: {nn_cv_error[model_num-1]:.4f}\")\n",
    "print(f\"Test Set Classification Error: {nn_test_error:.4f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Wrap Up\n",
    "\n",
    "In this lab, you practiced evaluating a model's performance and choosing between different model configurations. You split your datasets into training, cross validation, and test sets and saw how each of these are used in machine learning applications. In the next section of the course, you will see more tips on how to improve your models by diagnosing bias and variance. Keep it up!"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "provenance": []
  },
  "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}