notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "# Timeseries Cycle Predictions\n",
    "\n",
    "The goal of this notebook is to show how regression models can be used to predict the next \"cycle\" within a timeseries dataset.\n",
    "The data being used has been generated randomly, but you are free to modify the `generate_synthetic_data` function or incorporate your own data set.\n"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "abfac50bc59e289e"
  },
  {
   "cell_type": "code",
   "execution_count": 481,
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from datetime import timedelta"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:01.374223Z",
     "start_time": "2023-08-24T01:44:01.338516Z"
    }
   },
   "id": "714346af7184c61f"
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Generate synthetic data"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "1c1b8c9ab7a70ae4"
  },
  {
   "cell_type": "code",
   "execution_count": 482,
   "outputs": [],
   "source": [
    "# Seed\n",
    "np.random.seed(42)\n",
    "\n",
    "def generate_synthetic_data(num_users=10000):\n",
    "    user_ids = []\n",
    "    start_dates = []\n",
    "    end_dates = []\n",
    "    \n",
    "    # Generate random user IDs\n",
    "    unique_user_ids = np.random.choice(range(100000, 999999), num_users, replace=False)\n",
    "    # Generate random number of cycles for each user between 10 and 50\n",
    "    cycles_per_user = np.random.randint(10, 50, num_users)\n",
    "    start_date = pd.to_datetime('2023-01-01')\n",
    "    \n",
    "    for idx, user in enumerate(unique_user_ids):\n",
    "        cycle_start = start_date + timedelta(days=np.random.randint(7, 30))\n",
    "        for cycle in range(cycles_per_user[idx]):\n",
    "            # Generate a random start date for the first cycle and then subsequent cycles\n",
    "            if cycle != 0:\n",
    "                cycle_start += timedelta(days=np.random.randint(21, 35))  # Random gap between cycles\n",
    "\n",
    "            # Generate end date based on start date and random cycle length (let's say between 4 to 21 days)\n",
    "            cycle_end = cycle_start + timedelta(days=np.random.randint(4, 21))\n",
    "            \n",
    "            user_ids.append(user)\n",
    "            start_dates.append(cycle_start)\n",
    "            end_dates.append(cycle_end)\n",
    "            \n",
    "    return pd.DataFrame({\n",
    "        'USER_ID': user_ids,\n",
    "        'START_TIME': start_dates,\n",
    "        'END_TIME': end_dates\n",
    "    })\n"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:01.377952Z",
     "start_time": "2023-08-24T01:44:01.344104Z"
    }
   },
   "id": "44345a05a5c2a8df"
  },
  {
   "cell_type": "code",
   "execution_count": 483,
   "outputs": [
    {
     "data": {
      "text/plain": "   USER_ID START_TIME   END_TIME\n0   924197 2023-01-25 2023-02-04\n1   924197 2023-02-16 2023-02-26\n2   924197 2023-03-14 2023-03-25\n3   924197 2023-04-15 2023-04-30\n4   924197 2023-05-07 2023-05-20",
      "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>USER_ID</th>\n      <th>START_TIME</th>\n      <th>END_TIME</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>924197</td>\n      <td>2023-01-25</td>\n      <td>2023-02-04</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>924197</td>\n      <td>2023-02-16</td>\n      <td>2023-02-26</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>924197</td>\n      <td>2023-03-14</td>\n      <td>2023-03-25</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>924197</td>\n      <td>2023-04-15</td>\n      <td>2023-04-30</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>924197</td>\n      <td>2023-05-07</td>\n      <td>2023-05-20</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 483,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = generate_synthetic_data()\n",
    "df.head()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.734645Z",
     "start_time": "2023-08-24T01:44:01.389958Z"
    }
   },
   "id": "40f269bdf170a4ac"
  },
  {
   "cell_type": "code",
   "execution_count": 484,
   "outputs": [
    {
     "data": {
      "text/plain": "10000"
     },
     "execution_count": 484,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['USER_ID'].nunique()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.738368Z",
     "start_time": "2023-08-24T01:44:04.733504Z"
    }
   },
   "id": "5905d1ac8976ca6a"
  },
  {
   "cell_type": "code",
   "execution_count": 485,
   "outputs": [],
   "source": [
    "# Add day delta between start_time and end_time\n",
    "df['DURATION_DAYS'] = (df['END_TIME'] - df['START_TIME']).dt.days"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.743363Z",
     "start_time": "2023-08-24T01:44:04.738249Z"
    }
   },
   "id": "8e5fb947a5ab84f6"
  },
  {
   "cell_type": "code",
   "execution_count": 486,
   "outputs": [
    {
     "data": {
      "text/plain": "   USER_ID START_TIME   END_TIME  DURATION_DAYS\n0   924197 2023-01-25 2023-02-04             10\n1   924197 2023-02-16 2023-02-26             10\n2   924197 2023-03-14 2023-03-25             11\n3   924197 2023-04-15 2023-04-30             15\n4   924197 2023-05-07 2023-05-20             13",
      "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>USER_ID</th>\n      <th>START_TIME</th>\n      <th>END_TIME</th>\n      <th>DURATION_DAYS</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>924197</td>\n      <td>2023-01-25</td>\n      <td>2023-02-04</td>\n      <td>10</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>924197</td>\n      <td>2023-02-16</td>\n      <td>2023-02-26</td>\n      <td>10</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>924197</td>\n      <td>2023-03-14</td>\n      <td>2023-03-25</td>\n      <td>11</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>924197</td>\n      <td>2023-04-15</td>\n      <td>2023-04-30</td>\n      <td>15</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>924197</td>\n      <td>2023-05-07</td>\n      <td>2023-05-20</td>\n      <td>13</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 486,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.750514Z",
     "start_time": "2023-08-24T01:44:04.745481Z"
    }
   },
   "id": "d717d2f98fdfca6f"
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Exploratory Data Analysis"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "b39db0f661f1c884"
  },
  {
   "cell_type": "code",
   "execution_count": 487,
   "outputs": [],
   "source": [
    "import seaborn as sns"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.750616Z",
     "start_time": "2023-08-24T01:44:04.748132Z"
    }
   },
   "id": "91757ea7863cc0a1"
  },
  {
   "cell_type": "code",
   "execution_count": 488,
   "outputs": [
    {
     "data": {
      "text/plain": "DURATION_DAYS\n17    17480\n9     17437\n14    17391\n5     17390\n7     17324\n15    17308\n10    17248\n4     17203\n19    17183\n16    17153\n12    17147\n11    17139\n8     17130\n6     17112\n13    17109\n18    17077\n20    16935\nName: count, dtype: int64"
     },
     "execution_count": 488,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['DURATION_DAYS'].value_counts()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.767317Z",
     "start_time": "2023-08-24T01:44:04.752173Z"
    }
   },
   "id": "50532509e7c2db18"
  },
  {
   "cell_type": "code",
   "execution_count": 489,
   "outputs": [
    {
     "data": {
      "text/plain": "<Axes: xlabel='DURATION_DAYS', ylabel='Count'>"
     },
     "execution_count": 489,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.histplot(df['DURATION_DAYS'], kde=False, bins=30)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.942282Z",
     "start_time": "2023-08-24T01:44:04.755154Z"
    }
   },
   "id": "6424bb5d2faa85f5"
  },
  {
   "cell_type": "code",
   "execution_count": 490,
   "outputs": [
    {
     "data": {
      "text/plain": "       DURATION_DAYS\ncount  292766.000000\nmean       11.989541\nstd         4.894605\nmin         4.000000\n25%         8.000000\n50%        12.000000\n75%        16.000000\nmax        20.000000",
      "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>DURATION_DAYS</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>292766.000000</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>11.989541</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>4.894605</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>4.000000</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>8.000000</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>12.000000</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>16.000000</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>20.000000</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 490,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[['DURATION_DAYS']].describe()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:04.955041Z",
     "start_time": "2023-08-24T01:44:04.943636Z"
    }
   },
   "id": "ec60bc35928b08ca"
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Feature Engineering\n",
    "In this section, I am adding a simple feature for \"day of week\", such as weekend or weekday. "
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "f4e84e5806d6b000"
  },
  {
   "cell_type": "code",
   "execution_count": 491,
   "outputs": [
    {
     "data": {
      "text/plain": "   USER_ID START_TIME   END_TIME  DURATION_DAYS  START_TIME_DAY_TYPE  \\\n0   924197 2023-01-25 2023-02-04             10                    0   \n1   924197 2023-02-16 2023-02-26             10                    0   \n2   924197 2023-03-14 2023-03-25             11                    0   \n3   924197 2023-04-15 2023-04-30             15                    1   \n4   924197 2023-05-07 2023-05-20             13                    1   \n\n   END_TIME_DAY_TYPE PREV_CYCLE_END_TIME  DAYS_SINCE_LAST_CYCLE  GAP_TO_NEXT  \n0                  1                 NaT                    NaN         12.0  \n1                  1          2023-02-04                   12.0         16.0  \n2                  1          2023-02-26                   16.0         21.0  \n3                  1          2023-03-25                   21.0          7.0  \n4                  1          2023-04-30                    7.0         19.0  ",
      "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>USER_ID</th>\n      <th>START_TIME</th>\n      <th>END_TIME</th>\n      <th>DURATION_DAYS</th>\n      <th>START_TIME_DAY_TYPE</th>\n      <th>END_TIME_DAY_TYPE</th>\n      <th>PREV_CYCLE_END_TIME</th>\n      <th>DAYS_SINCE_LAST_CYCLE</th>\n      <th>GAP_TO_NEXT</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>924197</td>\n      <td>2023-01-25</td>\n      <td>2023-02-04</td>\n      <td>10</td>\n      <td>0</td>\n      <td>1</td>\n      <td>NaT</td>\n      <td>NaN</td>\n      <td>12.0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>924197</td>\n      <td>2023-02-16</td>\n      <td>2023-02-26</td>\n      <td>10</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2023-02-04</td>\n      <td>12.0</td>\n      <td>16.0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>924197</td>\n      <td>2023-03-14</td>\n      <td>2023-03-25</td>\n      <td>11</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2023-02-26</td>\n      <td>16.0</td>\n      <td>21.0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>924197</td>\n      <td>2023-04-15</td>\n      <td>2023-04-30</td>\n      <td>15</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2023-03-25</td>\n      <td>21.0</td>\n      <td>7.0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>924197</td>\n      <td>2023-05-07</td>\n      <td>2023-05-20</td>\n      <td>13</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2023-04-30</td>\n      <td>7.0</td>\n      <td>19.0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 491,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['START_TIME_DAY_TYPE'] = df['START_TIME'].dt.dayofweek.apply(lambda x: 0 if x < 5 else 1)\n",
    "df['END_TIME_DAY_TYPE'] = df['END_TIME'].dt.dayofweek.apply(lambda x: 0 if x < 5 else 1)\n",
    "\n",
    "df['PREV_CYCLE_END_TIME'] = df.groupby('USER_ID')['END_TIME'].shift(1)\n",
    "df['DAYS_SINCE_LAST_CYCLE'] = abs(df['PREV_CYCLE_END_TIME'] - df['START_TIME']).dt.days\n",
    "df['GAP_TO_NEXT'] = df['DAYS_SINCE_LAST_CYCLE'].shift(-1)\n",
    "\n",
    "df.head()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:05.082150Z",
     "start_time": "2023-08-24T01:44:04.954401Z"
    }
   },
   "id": "74895ecbc0001a7d"
  },
  {
   "cell_type": "code",
   "execution_count": 492,
   "outputs": [
    {
     "data": {
      "text/plain": "<Axes: xlabel='DAYS_SINCE_LAST_CYCLE', ylabel='Count'>"
     },
     "execution_count": 492,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.histplot(data=df, x=df['DAYS_SINCE_LAST_CYCLE'], kde=False)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:05.273389Z",
     "start_time": "2023-08-24T01:44:05.080369Z"
    }
   },
   "id": "37dad7c9569c6524"
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Predict next cycle start date"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "99b623f6cfc9ea78"
  },
  {
   "cell_type": "code",
   "execution_count": 493,
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "from sklearn.metrics import mean_absolute_error"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:05.278068Z",
     "start_time": "2023-08-24T01:44:05.273490Z"
    }
   },
   "id": "5043303d2fb88c6e"
  },
  {
   "cell_type": "markdown",
   "source": [
    "### Make it production-ready"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "c799d4e9ac6a6442"
  },
  {
   "cell_type": "code",
   "execution_count": 494,
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "def create_and_save_model(df, filename=\"model/model_cycle_prediction.pk\"):\n",
    "    # Remove rows with NaN values in the target column\n",
    "    df.dropna(subset=['GAP_TO_NEXT', 'DAYS_SINCE_LAST_CYCLE', 'PREV_CYCLE_END_TIME'], inplace=True)\n",
    "\n",
    "    # Features and target\n",
    "    X = df[['DURATION_DAYS', 'START_TIME_DAY_TYPE', 'END_TIME_DAY_TYPE', 'DAYS_SINCE_LAST_CYCLE']]\n",
    "    y = df['GAP_TO_NEXT']\n",
    "\n",
    "    # Split data\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)\n",
    "\n",
    "    # Initialize and train model\n",
    "    model = GradientBoostingRegressor(n_estimators=100, random_state=42)\n",
    "    model.fit(X_train, y_train)\n",
    "\n",
    "    # Predict on test set\n",
    "    y_pred = model.predict(X_test)\n",
    "\n",
    "    # Calculate error\n",
    "    error = mean_absolute_error(y_test, y_pred)\n",
    "    print(f\"Mean Absolute Error in predicting gap to next cycle start: {error} days, or {int(error)} days rounded\")\n",
    "\n",
    "    # Pickle the trained model\n",
    "    with open(filename, 'wb') as file:\n",
    "        pickle.dump(model, file)\n",
    "\n",
    "    return filename\n"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:05.283205Z",
     "start_time": "2023-08-24T01:44:05.278145Z"
    }
   },
   "id": "72cbc50034e5287c"
  },
  {
   "cell_type": "code",
   "execution_count": 495,
   "outputs": [],
   "source": [
    "def predict_next_cycle_start(user_id, model_path, df):\n",
    "    # Load the pickled model\n",
    "    with open(model_path, 'rb') as file:\n",
    "        model = pickle.load(file)\n",
    "\n",
    "    # Extract the most recent cycle data for the user\n",
    "    user_data = df[df['USER_ID'] == user_id]\n",
    "    most_recent_data = user_data.iloc[-1]\n",
    "\n",
    "    # Prepare this data in the same format as used for training\n",
    "    X_single = most_recent_data[['DURATION_DAYS', 'START_TIME_DAY_TYPE', 'END_TIME_DAY_TYPE', 'DAYS_SINCE_LAST_CYCLE']]\n",
    "\n",
    "    # Predict the gap to the next cycle start\n",
    "    predicted_gap = model.predict([X_single])\n",
    "\n",
    "    # Calculate the predicted start date for the next cycle\n",
    "    predicted_start_date = pd.to_datetime(most_recent_data['END_TIME']) + pd.Timedelta(days=int(predicted_gap[0]))\n",
    "\n",
    "    return predicted_start_date\n"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:05.283320Z",
     "start_time": "2023-08-24T01:44:05.280796Z"
    }
   },
   "id": "1b95d688825ccf76"
  },
  {
   "cell_type": "markdown",
   "source": [
    "## Test model"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "8f5d3d6201851d31"
  },
  {
   "cell_type": "code",
   "execution_count": 496,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean Absolute Error in predicting gap to next cycle start: 3.4903614351812156 days, or 3 days rounded\n"
     ]
    }
   ],
   "source": [
    "model_path = create_and_save_model(df)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:09.754522Z",
     "start_time": "2023-08-24T01:44:05.284708Z"
    }
   },
   "id": "67d48dedad099dd1"
  },
  {
   "cell_type": "code",
   "execution_count": 497,
   "outputs": [
    {
     "data": {
      "text/plain": "219504"
     },
     "execution_count": 497,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pick a user\n",
    "random_user = df.sample()['USER_ID'].values[0]\n",
    "random_user"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:09.761176Z",
     "start_time": "2023-08-24T01:44:09.756819Z"
    }
   },
   "id": "cb0a745146d6434b"
  },
  {
   "cell_type": "code",
   "execution_count": 498,
   "outputs": [
    {
     "data": {
      "text/plain": "       USER_ID START_TIME   END_TIME  DURATION_DAYS  START_TIME_DAY_TYPE  \\\n95415   219504 2023-02-19 2023-03-04             13                    1   \n95416   219504 2023-03-19 2023-04-07             19                    1   \n95417   219504 2023-04-12 2023-04-22             10                    0   \n95418   219504 2023-05-03 2023-05-15             12                    0   \n95419   219504 2023-05-25 2023-06-13             19                    0   \n95420   219504 2023-06-20 2023-06-29              9                    0   \n95421   219504 2023-07-19 2023-08-05             17                    0   \n95422   219504 2023-08-22 2023-09-01             10                    0   \n95423   219504 2023-09-16 2023-09-22              6                    1   \n95424   219504 2023-10-09 2023-10-15              6                    0   \n95425   219504 2023-10-31 2023-11-18             18                    0   \n95426   219504 2023-11-23 2023-12-08             15                    0   \n95427   219504 2023-12-23 2024-01-05             13                    1   \n95428   219504 2024-01-16 2024-01-23              7                    0   \n95429   219504 2024-02-12 2024-02-17              5                    0   \n95430   219504 2024-03-16 2024-03-27             11                    1   \n95431   219504 2024-04-10 2024-04-19              9                    0   \n95432   219504 2024-05-11 2024-05-30             19                    1   \n95433   219504 2024-06-07 2024-06-17             10                    0   \n95434   219504 2024-07-05 2024-07-22             17                    0   \n95435   219504 2024-08-02 2024-08-21             19                    0   \n95436   219504 2024-08-27 2024-09-02              6                    0   \n95437   219504 2024-09-20 2024-09-26              6                    0   \n95438   219504 2024-10-21 2024-10-26              5                    0   \n95439   219504 2024-11-19 2024-11-27              8                    0   \n\n       END_TIME_DAY_TYPE PREV_CYCLE_END_TIME  DAYS_SINCE_LAST_CYCLE  \\\n95415                  1          2023-02-02                   17.0   \n95416                  0          2023-03-04                   15.0   \n95417                  1          2023-04-07                    5.0   \n95418                  0          2023-04-22                   11.0   \n95419                  0          2023-05-15                   10.0   \n95420                  0          2023-06-13                    7.0   \n95421                  1          2023-06-29                   20.0   \n95422                  0          2023-08-05                   17.0   \n95423                  0          2023-09-01                   15.0   \n95424                  1          2023-09-22                   17.0   \n95425                  1          2023-10-15                   16.0   \n95426                  0          2023-11-18                    5.0   \n95427                  0          2023-12-08                   15.0   \n95428                  0          2024-01-05                   11.0   \n95429                  1          2024-01-23                   20.0   \n95430                  0          2024-02-17                   28.0   \n95431                  0          2024-03-27                   14.0   \n95432                  0          2024-04-19                   22.0   \n95433                  0          2024-05-30                    8.0   \n95434                  0          2024-06-17                   18.0   \n95435                  0          2024-07-22                   11.0   \n95436                  0          2024-08-21                    6.0   \n95437                  0          2024-09-02                   18.0   \n95438                  1          2024-09-26                   25.0   \n95439                  0          2024-10-26                   24.0   \n\n       GAP_TO_NEXT  \n95415         15.0  \n95416          5.0  \n95417         11.0  \n95418         10.0  \n95419          7.0  \n95420         20.0  \n95421         17.0  \n95422         15.0  \n95423         17.0  \n95424         16.0  \n95425          5.0  \n95426         15.0  \n95427         11.0  \n95428         20.0  \n95429         28.0  \n95430         14.0  \n95431         22.0  \n95432          8.0  \n95433         18.0  \n95434         11.0  \n95435          6.0  \n95436         18.0  \n95437         25.0  \n95438         24.0  \n95439         16.0  ",
      "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>USER_ID</th>\n      <th>START_TIME</th>\n      <th>END_TIME</th>\n      <th>DURATION_DAYS</th>\n      <th>START_TIME_DAY_TYPE</th>\n      <th>END_TIME_DAY_TYPE</th>\n      <th>PREV_CYCLE_END_TIME</th>\n      <th>DAYS_SINCE_LAST_CYCLE</th>\n      <th>GAP_TO_NEXT</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>95415</th>\n      <td>219504</td>\n      <td>2023-02-19</td>\n      <td>2023-03-04</td>\n      <td>13</td>\n      <td>1</td>\n      <td>1</td>\n      <td>2023-02-02</td>\n      <td>17.0</td>\n      <td>15.0</td>\n    </tr>\n    <tr>\n      <th>95416</th>\n      <td>219504</td>\n      <td>2023-03-19</td>\n      <td>2023-04-07</td>\n      <td>19</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2023-03-04</td>\n      <td>15.0</td>\n      <td>5.0</td>\n    </tr>\n    <tr>\n      <th>95417</th>\n      <td>219504</td>\n      <td>2023-04-12</td>\n      <td>2023-04-22</td>\n      <td>10</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2023-04-07</td>\n      <td>5.0</td>\n      <td>11.0</td>\n    </tr>\n    <tr>\n      <th>95418</th>\n      <td>219504</td>\n      <td>2023-05-03</td>\n      <td>2023-05-15</td>\n      <td>12</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2023-04-22</td>\n      <td>11.0</td>\n      <td>10.0</td>\n    </tr>\n    <tr>\n      <th>95419</th>\n      <td>219504</td>\n      <td>2023-05-25</td>\n      <td>2023-06-13</td>\n      <td>19</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2023-05-15</td>\n      <td>10.0</td>\n      <td>7.0</td>\n    </tr>\n    <tr>\n      <th>95420</th>\n      <td>219504</td>\n      <td>2023-06-20</td>\n      <td>2023-06-29</td>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2023-06-13</td>\n      <td>7.0</td>\n      <td>20.0</td>\n    </tr>\n    <tr>\n      <th>95421</th>\n      <td>219504</td>\n      <td>2023-07-19</td>\n      <td>2023-08-05</td>\n      <td>17</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2023-06-29</td>\n      <td>20.0</td>\n      <td>17.0</td>\n    </tr>\n    <tr>\n      <th>95422</th>\n      <td>219504</td>\n      <td>2023-08-22</td>\n      <td>2023-09-01</td>\n      <td>10</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2023-08-05</td>\n      <td>17.0</td>\n      <td>15.0</td>\n    </tr>\n    <tr>\n      <th>95423</th>\n      <td>219504</td>\n      <td>2023-09-16</td>\n      <td>2023-09-22</td>\n      <td>6</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2023-09-01</td>\n      <td>15.0</td>\n      <td>17.0</td>\n    </tr>\n    <tr>\n      <th>95424</th>\n      <td>219504</td>\n      <td>2023-10-09</td>\n      <td>2023-10-15</td>\n      <td>6</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2023-09-22</td>\n      <td>17.0</td>\n      <td>16.0</td>\n    </tr>\n    <tr>\n      <th>95425</th>\n      <td>219504</td>\n      <td>2023-10-31</td>\n      <td>2023-11-18</td>\n      <td>18</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2023-10-15</td>\n      <td>16.0</td>\n      <td>5.0</td>\n    </tr>\n    <tr>\n      <th>95426</th>\n      <td>219504</td>\n      <td>2023-11-23</td>\n      <td>2023-12-08</td>\n      <td>15</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2023-11-18</td>\n      <td>5.0</td>\n      <td>15.0</td>\n    </tr>\n    <tr>\n      <th>95427</th>\n      <td>219504</td>\n      <td>2023-12-23</td>\n      <td>2024-01-05</td>\n      <td>13</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2023-12-08</td>\n      <td>15.0</td>\n      <td>11.0</td>\n    </tr>\n    <tr>\n      <th>95428</th>\n      <td>219504</td>\n      <td>2024-01-16</td>\n      <td>2024-01-23</td>\n      <td>7</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-01-05</td>\n      <td>11.0</td>\n      <td>20.0</td>\n    </tr>\n    <tr>\n      <th>95429</th>\n      <td>219504</td>\n      <td>2024-02-12</td>\n      <td>2024-02-17</td>\n      <td>5</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2024-01-23</td>\n      <td>20.0</td>\n      <td>28.0</td>\n    </tr>\n    <tr>\n      <th>95430</th>\n      <td>219504</td>\n      <td>2024-03-16</td>\n      <td>2024-03-27</td>\n      <td>11</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2024-02-17</td>\n      <td>28.0</td>\n      <td>14.0</td>\n    </tr>\n    <tr>\n      <th>95431</th>\n      <td>219504</td>\n      <td>2024-04-10</td>\n      <td>2024-04-19</td>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-03-27</td>\n      <td>14.0</td>\n      <td>22.0</td>\n    </tr>\n    <tr>\n      <th>95432</th>\n      <td>219504</td>\n      <td>2024-05-11</td>\n      <td>2024-05-30</td>\n      <td>19</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2024-04-19</td>\n      <td>22.0</td>\n      <td>8.0</td>\n    </tr>\n    <tr>\n      <th>95433</th>\n      <td>219504</td>\n      <td>2024-06-07</td>\n      <td>2024-06-17</td>\n      <td>10</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-05-30</td>\n      <td>8.0</td>\n      <td>18.0</td>\n    </tr>\n    <tr>\n      <th>95434</th>\n      <td>219504</td>\n      <td>2024-07-05</td>\n      <td>2024-07-22</td>\n      <td>17</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-06-17</td>\n      <td>18.0</td>\n      <td>11.0</td>\n    </tr>\n    <tr>\n      <th>95435</th>\n      <td>219504</td>\n      <td>2024-08-02</td>\n      <td>2024-08-21</td>\n      <td>19</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-07-22</td>\n      <td>11.0</td>\n      <td>6.0</td>\n    </tr>\n    <tr>\n      <th>95436</th>\n      <td>219504</td>\n      <td>2024-08-27</td>\n      <td>2024-09-02</td>\n      <td>6</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-08-21</td>\n      <td>6.0</td>\n      <td>18.0</td>\n    </tr>\n    <tr>\n      <th>95437</th>\n      <td>219504</td>\n      <td>2024-09-20</td>\n      <td>2024-09-26</td>\n      <td>6</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-09-02</td>\n      <td>18.0</td>\n      <td>25.0</td>\n    </tr>\n    <tr>\n      <th>95438</th>\n      <td>219504</td>\n      <td>2024-10-21</td>\n      <td>2024-10-26</td>\n      <td>5</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2024-09-26</td>\n      <td>25.0</td>\n      <td>24.0</td>\n    </tr>\n    <tr>\n      <th>95439</th>\n      <td>219504</td>\n      <td>2024-11-19</td>\n      <td>2024-11-27</td>\n      <td>8</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2024-10-26</td>\n      <td>24.0</td>\n      <td>16.0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 498,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['USER_ID'] == random_user]"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:09.769250Z",
     "start_time": "2023-08-24T01:44:09.760063Z"
    }
   },
   "id": "e507197a512c9cd3"
  },
  {
   "cell_type": "code",
   "execution_count": 499,
   "outputs": [
    {
     "data": {
      "text/plain": "USER_ID                               219504\nSTART_TIME               2024-11-19 00:00:00\nEND_TIME                 2024-11-27 00:00:00\nDURATION_DAYS                              8\nSTART_TIME_DAY_TYPE                        0\nEND_TIME_DAY_TYPE                          0\nPREV_CYCLE_END_TIME      2024-10-26 00:00:00\nDAYS_SINCE_LAST_CYCLE                   24.0\nGAP_TO_NEXT                             16.0\nName: 95439, dtype: object"
     },
     "execution_count": 499,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['USER_ID'] == random_user].iloc[-1]"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:09.772742Z",
     "start_time": "2023-08-24T01:44:09.768204Z"
    }
   },
   "id": "80417d8a79ddac3f"
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "user_id = random_user  # Replace with the desired user's ID\n",
    "predicted_date = predict_next_cycle_start(user_id, model_path, df)\n",
    "print(f\"Predicted start date of the next cycle for user {user_id}: {predicted_date.date()}\")"
   ],
   "metadata": {
    "collapsed": false
   },
   "id": "334a43f02676c289"
  },
  {
   "cell_type": "code",
   "execution_count": 501,
   "outputs": [],
   "source": [
    "# Save dataframe\n",
    "df.to_csv('data/users.csv', encoding='utf-8', index=False)"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2023-08-24T01:44:10.494257Z",
     "start_time": "2023-08-24T01:44:09.777803Z"
    }
   },
   "id": "eb52a312e3d2676a"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}