audiobook_data_preprocessing.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "audiobook-data-preprocessing.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "u7S-D3wNGDQQ"
      },
      "source": [
        "# **Audiobook Purchase Prediction**\n",
        "## **Project description**\n",
        "\n",
        "The given data comes from an Audiobook app. Each customer in the database has made a purchase at least once, that's why he/she is in the database. We want to create an algorithm based on our available data that can predict if a customer will buy again from the Audiobook company.\n",
        "\n",
        "The main idea is that if a customer has a low probability of coming back, there is no reason to spend any money on advertizing to him/her. If we can focus our efforts ONLY on customers that are likely to convert again, we can make great savings. Moreover, this model can identify the most important metrics for a customer to come back again. Identifying new customers creates value and growth opportunities.\n",
        "\n",
        "the data is summarized in a .csv file.\n",
        "\n",
        "### **Variables:**\n",
        "\n",
        "*   Customer ID\n",
        "*   Book length in mins_avg (average of all purchases)\n",
        "*   Book length in minutes_sum (sum of all purchases. If equal to Book length in mins_avg, thus customers made 1 purchase only.)\n",
        "*   Price Paid_avg (average of all purchases)\n",
        "*   Price paid_sum (sum of all purchases. If equal to Price Paid_avg, thus customers made 1 purchase only.)\n",
        "*   Review (Boolean variable. 1 = Customer left a review)\n",
        "*   Review (out of 10)\n",
        "*   Total minutes listened\n",
        "*   Completion (from 0 to 1) => Total minutes listened / Total lengths of books the person has purchased\n",
        "*   Support requests (number)\n",
        "*   Last visited minus purchase date (in days) => measures the difference between the last time a person interacted with the platform and the first purchase date. The bigger the difference, the bigger the engagement.\n",
        "\n",
        "So these variables are going to be the inputs of our model (excluding customer ID, as it is completely arbitrary).\n",
        "\n",
        "### **Targets:**\n",
        "\n",
        "The targets are a Boolean variable (0 or 1). We are taking a period of 2 years in our inputs, and the next 6 months as targets.<br />\n",
        "In fact, we are predicting if based on the last 2 years of activity and engagement, a customer will convert in the next 6 months. If they do not convert after 6 months, chances are they have gone to a competitor or di not like the Audiobook way of digesting information.\n",
        "\n",
        "**This is a supervised classification problem with two classes: won't buy and will buy, represented by 0s and 1s.**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z7_6FraUN3iY"
      },
      "source": [
        "# import required packages\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "from sklearn import preprocessing"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 198
        },
        "id": "YVEcyeBmOHN8",
        "outputId": "8ee88378-6f6f-4b72-f81a-8cfb1bc4566e"
      },
      "source": [
        "# import data\n",
        "column_names = [\"customer_id\", \"book_length_mins_avg\", \"book_length_mins_sum\",\n",
        "                \"price_paid_avg\", \"price_paid_sum\", \"left_review\", \"review\",\n",
        "                \"total_mins_listened\", \"completion\", \"support_requests\",\n",
        "                \"last_visited_minus_purchase_date\", \"label\"]\n",
        "\n",
        "audiobook_data = pd.read_csv(\"audiobooks-data.csv\", header=None)\n",
        "audiobook_data.columns = column_names\n",
        "audiobook_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>customer_id</th>\n",
              "      <th>book_length_mins_avg</th>\n",
              "      <th>book_length_mins_sum</th>\n",
              "      <th>price_paid_avg</th>\n",
              "      <th>price_paid_sum</th>\n",
              "      <th>left_review</th>\n",
              "      <th>review</th>\n",
              "      <th>total_mins_listened</th>\n",
              "      <th>completion</th>\n",
              "      <th>support_requests</th>\n",
              "      <th>last_visited_minus_purchase_date</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>994</td>\n",
              "      <td>1620.0</td>\n",
              "      <td>1620</td>\n",
              "      <td>19.73</td>\n",
              "      <td>19.73</td>\n",
              "      <td>1</td>\n",
              "      <td>10.00</td>\n",
              "      <td>0.99</td>\n",
              "      <td>1603.8</td>\n",
              "      <td>5</td>\n",
              "      <td>92</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1143</td>\n",
              "      <td>2160.0</td>\n",
              "      <td>2160</td>\n",
              "      <td>5.33</td>\n",
              "      <td>5.33</td>\n",
              "      <td>0</td>\n",
              "      <td>8.91</td>\n",
              "      <td>0.00</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2059</td>\n",
              "      <td>2160.0</td>\n",
              "      <td>2160</td>\n",
              "      <td>5.33</td>\n",
              "      <td>5.33</td>\n",
              "      <td>0</td>\n",
              "      <td>8.91</td>\n",
              "      <td>0.00</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0</td>\n",
              "      <td>388</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2882</td>\n",
              "      <td>1620.0</td>\n",
              "      <td>1620</td>\n",
              "      <td>5.96</td>\n",
              "      <td>5.96</td>\n",
              "      <td>0</td>\n",
              "      <td>8.91</td>\n",
              "      <td>0.42</td>\n",
              "      <td>680.4</td>\n",
              "      <td>1</td>\n",
              "      <td>129</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>3342</td>\n",
              "      <td>2160.0</td>\n",
              "      <td>2160</td>\n",
              "      <td>5.33</td>\n",
              "      <td>5.33</td>\n",
              "      <td>0</td>\n",
              "      <td>8.91</td>\n",
              "      <td>0.22</td>\n",
              "      <td>475.2</td>\n",
              "      <td>0</td>\n",
              "      <td>361</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   customer_id  book_length_mins_avg  ...  last_visited_minus_purchase_date  label\n",
              "0          994                1620.0  ...                                92      0\n",
              "1         1143                2160.0  ...                                 0      0\n",
              "2         2059                2160.0  ...                               388      0\n",
              "3         2882                1620.0  ...                               129      0\n",
              "4         3342                2160.0  ...                               361      0\n",
              "\n",
              "[5 rows x 12 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PJj-oubjQ-v8",
        "outputId": "fa032090-b601-4017-cf3a-3f30cc33edd4"
      },
      "source": [
        "if audiobook_data.shape[0] == audiobook_data[\"customer_id\"].nunique():\n",
        "  print(f\"The Audiobook data contains information about {audiobook_data.shape[0]} customers.\")\n",
        "else:\n",
        "  print(\"Check for duplicates in customer_id column.\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "The Audiobook data contains information about 14084 customers.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "j-BDy_LxS1Qh",
        "outputId": "e4b897a3-d609-4f82-8bb5-20cebcf1684f"
      },
      "source": [
        "print(audiobook_data.info())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 14084 entries, 0 to 14083\n",
            "Data columns (total 12 columns):\n",
            " #   Column                            Non-Null Count  Dtype  \n",
            "---  ------                            --------------  -----  \n",
            " 0   customer_id                       14084 non-null  int64  \n",
            " 1   book_length_mins_avg              14084 non-null  float64\n",
            " 2   book_length_mins_sum              14084 non-null  int64  \n",
            " 3   price_paid_avg                    14084 non-null  float64\n",
            " 4   price_paid_sum                    14084 non-null  float64\n",
            " 5   left_review                       14084 non-null  int64  \n",
            " 6   review                            14084 non-null  float64\n",
            " 7   total_mins_listened               14084 non-null  float64\n",
            " 8   completion                        14084 non-null  float64\n",
            " 9   support_requests                  14084 non-null  int64  \n",
            " 10  last_visited_minus_purchase_date  14084 non-null  int64  \n",
            " 11  label                             14084 non-null  int64  \n",
            "dtypes: float64(6), int64(6)\n",
            "memory usage: 1.3 MB\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 424
        },
        "id": "8RkV3Fsyax0q",
        "outputId": "bd73765c-c04e-4131-d344-319bdb57dcf9"
      },
      "source": [
        "print(f\"Label distribution (in %):\\n{round(audiobook_data['label'].value_counts() * 100 / audiobook_data.shape[0], 2)}\")\n",
        "\n",
        "# plot label distribution\n",
        "plt.figure(figsize=(10,5))\n",
        "sns.countplot(x=\"label\", data=audiobook_data)\n",
        "plt.title(\"Label distribution\", size=14)\n",
        "sns.despine()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Label distribution (in %):\n",
            "0    84.12\n",
            "1    15.88\n",
            "Name: label, dtype: float64\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        },
        "id": "w9qKDbyRYTcO",
        "outputId": "522dfcb9-cfe8-4965-d385-e5374e0f2203"
      },
      "source": [
        "fig, axes = plt.subplots(1, 2, figsize=(16, 5), sharey=False, sharex=False)\n",
        "sns.kdeplot(x=\"book_length_mins_avg\", data=audiobook_data, hue=\"label\", shade=True, ax=axes[0])\n",
        "sns.kdeplot(x=\"price_paid_avg\", data=audiobook_data, hue=\"label\", shade=True, ax=axes[1])\n",
        "axes[0].set_title(\"Distribution of average of book length per label (in minutes)\")\n",
        "axes[0].set_xlabel(\"Average book length (in minutes)\")\n",
        "axes[1].set_title(\"Distribution of average price paid per label\")\n",
        "axes[1].set_xlabel(\"Average price paid\")\n",
        "axes[1].set_xlim(0, 30)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x360 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jOXuH9zXbwW4"
      },
      "source": [
        "Average Book Length and Average Price Paid for both labels are following the same distribution pattern. Nevertheless, the dataset is clearly unbalanced, as label 0 represents 84% of our dataset. This could affect the behavior of our future model. We need to balance the dataset consequently."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tU7nrDBYOOU2"
      },
      "source": [
        "## Balancing the dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aD_ZEcxfN9Mn"
      },
      "source": [
        "# store input data into a numpy array excluding cutomer_id and label columns\n",
        "unscaled_inputs = audiobook_data.iloc[:, 1:-1].values\n",
        "\n",
        "# store target data into a separate numpy array\n",
        "targets = audiobook_data.iloc[:, -1].values"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rZflee6PX9XY"
      },
      "source": [
        "# When the data was collected it was actually arranged by date\n",
        "# Shuffle the indices of the data, so the data is not arranged in any way when we feed it.\n",
        "# Since we will be batching, we want the data to be as randomly spread out as possible\n",
        "shuffled_indices = np.arange(unscaled_inputs.shape[0])\n",
        "np.random.shuffle(shuffled_indices)\n",
        "\n",
        "# Use the shuffled indices to shuffle the inputs and targets.\n",
        "unscaled_inputs = unscaled_inputs[shuffled_indices]\n",
        "targets = targets[shuffled_indices]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0zUzqYl5Sote"
      },
      "source": [
        "We are going to keep as many 0s as 1s and delete the others to get a balanced dataset."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PrmaIYqqRztQ"
      },
      "source": [
        "# count number of 1s in targets\n",
        "num_of_ones = int(np.sum(targets))\n",
        "zero_targets_counter = 0\n",
        "indices_to_remove = []\n",
        "\n",
        "# once number of 0s reached number of 1s, we delete remaining rows with label 0\n",
        "for i in range(targets.shape[0]):\n",
        "  if targets[i] == 0:\n",
        "    zero_targets_counter += 1\n",
        "    if zero_targets_counter > num_of_ones:\n",
        "      indices_to_remove.append(i)\n",
        "\n",
        "unscaled_inputs_equal_priors = np.delete(unscaled_inputs, indices_to_remove, axis=0)\n",
        "targets_equal_priors = np.delete(targets, indices_to_remove, axis=0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UFJU5PGvUMP_"
      },
      "source": [
        "## Standardize the inputs"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XrESSrphUDX4"
      },
      "source": [
        "# center to the mean and scale to unit variance\n",
        "scaled_inputs = preprocessing.scale(unscaled_inputs_equal_priors)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AS6OZ0cpUZat"
      },
      "source": [
        "## Shuffle the data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ybj-B0VqUYkO"
      },
      "source": [
        "# shuffle data again to avoid that all 1s are stored in train set\n",
        "shuffled_indices = np.arange(scaled_inputs.shape[0])\n",
        "np.random.shuffle(shuffled_indices)\n",
        "\n",
        "shuffled_inputs = scaled_inputs[shuffled_indices]\n",
        "shuffled_targets = targets_equal_priors[shuffled_indices]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zK58XFX5U8dh"
      },
      "source": [
        "## Split the dataset into train, validation and test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yODph_7AU6hF",
        "outputId": "b9e3b945-c122-42ab-eaeb-93992be3f488"
      },
      "source": [
        "sample_count = shuffled_inputs.shape[0]\n",
        "\n",
        "# define size of train, validation and test sets\n",
        "train_samples_count = int(0.8 * sample_count)\n",
        "validation_samples_count = int(0.1 * sample_count)\n",
        "test_samples_count = sample_count - train_samples_count - validation_samples_count\n",
        "\n",
        "# build train set with shuffled data\n",
        "train_inputs = shuffled_inputs[:train_samples_count]\n",
        "train_targets = shuffled_targets[:train_samples_count]\n",
        "\n",
        "# build validation set with shuffled data\n",
        "validation_inputs = shuffled_inputs[train_samples_count:train_samples_count + validation_samples_count]\n",
        "validation_targets = shuffled_targets[train_samples_count:train_samples_count + validation_samples_count]\n",
        "\n",
        "# build test set with shuffled data\n",
        "test_inputs = shuffled_inputs[train_samples_count + validation_samples_count:]\n",
        "test_targets = shuffled_targets[train_samples_count + validation_samples_count:]\n",
        "\n",
        "# print label distribution\n",
        "print(f\"Train set:\\nNumber of 1s: {np.sum(train_targets)} | Set size: {train_samples_count} | % of 1s: {np.sum(train_targets) / train_samples_count}\")\n",
        "print(\"-------------\")\n",
        "print(f\"Validation set:\\nNumber of 1s: {np.sum(validation_targets)} | Set size: {validation_samples_count} | % of 1s: {np.sum(validation_targets) / validation_samples_count}\")\n",
        "print(\"-------------\")\n",
        "print(f\"Test set:\\nNumber of 1s: {np.sum(test_targets)} | Set size: {test_samples_count} | % of 1s: {np.sum(test_targets) / test_samples_count}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train set:\n",
            "Number of 1s: 1765 | Set size: 3579 | % of 1s: 0.49315451243364067\n",
            "-------------\n",
            "Validation set:\n",
            "Number of 1s: 229 | Set size: 447 | % of 1s: 0.5123042505592841\n",
            "-------------\n",
            "Test set:\n",
            "Number of 1s: 243 | Set size: 448 | % of 1s: 0.5424107142857143\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jzk2GdWp7krZ"
      },
      "source": [
        "<p>The dataset is now pretty well balanced !</p>\n",
        "\n",
        "<p>Let's save the three datasets in .npz format to use them for our future algorithms.<br />\n",
        "Each time we run this notebook, we will process the data once again (forgetting the previous preprocessing), so training, validation and test datasets will contain different samples.</p>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HImdSN4dV55Y"
      },
      "source": [
        "# Save the three datasets in *.npz.\n",
        "np.savez(\"audiobooks-data-train\", inputs=train_inputs, targets=train_targets)\n",
        "np.savez(\"audiobooks-data-validation\", inputs=validation_inputs, targets=validation_targets)\n",
        "np.savez(\"audiobooks-data-test\", inputs=test_inputs, targets=test_targets)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "h-mkHH498Ue4"
      },
      "source": [
        "We are going to build a Deep Learning model in a separate notebook using these .npz files."
      ]
    }
  ]
}