bikeshare-modelling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "2e7efabc-3975-46b5-b5e2-b620a8e8785f",
   "metadata": {},
   "source": [
    "# Bikeshare Modelling\n",
    "\n",
    "Zach O'Brien"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "8c956cff-b856-40d3-9770-5416db6e8b2f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "np.random.seed(9)  # For reproducibility\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import sklearn\n",
    "from sklearn.model_selection import TimeSeriesSplit\n",
    "import sklearn.linear_model as linear_model\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.preprocessing import StandardScaler, MinMaxScaler"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d791520-f28b-4c02-abb0-c6372a0b5a1f",
   "metadata": {},
   "source": [
    "# 1. Data Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "942c4a29-9245-4eed-8b9c-e4abcca98fcc",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def preprocess_bikeshare_data(df, categorical_cols, label_col):\n",
    "    \"\"\"Preprocess the Bikeshare Ridership Data\n",
    "    \n",
    "    One-hot encodes categorical columns, and applies MinMaxScaler to the\n",
    "    label column \"cnt\" because it will be a feature as well as a label\n",
    "    in the time series models.\n",
    "    \n",
    "    Args:\n",
    "        df (pd.DataFrame):\n",
    "            Data\n",
    "        categorical_cols (list[str]):\n",
    "            Names of columns to one-hot encode\n",
    "        label_col (str):\n",
    "            Name of the label column (column with label)\n",
    "    \n",
    "    Returns:\n",
    "        pd.DataFrame, MinMaxScaler:\n",
    "            Data with categorical values one-hot encoded and labels scaled,\n",
    "            and the scaler fit to the original label range\n",
    "    \"\"\"\n",
    "    new_df = one_hot_encode(df, column_names=categorical_cols)\n",
    "    new_df, scaler = min_max_scale(new_df, col=label_col)\n",
    "    return new_df, scaler\n",
    "\n",
    "\n",
    "def one_hot_encode(df, column_names):\n",
    "    \"\"\"One-hot encode columns in a DataFrame\n",
    "\n",
    "    Args:\n",
    "        df (pd.DataFrame): \n",
    "            Data\n",
    "        column_names (list[str]): \n",
    "            Names of columns which should be converted to one-hot encoding\n",
    "\n",
    "    Returns:\n",
    "        pd.DataFrame:\n",
    "            Data with the specified columns converted to one-hot encoding\n",
    "    \"\"\"\n",
    "    df_copy = df.copy()\n",
    "    dummies = [pd.get_dummies(df_copy[col], prefix=col) for col in column_names]\n",
    "    df_copy.drop(column_names, axis=1, inplace=True)\n",
    "    return pd.concat(dummies + [df_copy], axis=1)\n",
    "\n",
    "\n",
    "def min_max_scale(df, col):\n",
    "    \"\"\"Apply MinMax scaling to one column in a DataFrame\n",
    "    \n",
    "    Args:\n",
    "        df (pd.DataFrame):\n",
    "            Data with a column of real values named `col`\n",
    "        col (str):\n",
    "            Name of the column to scale\n",
    "    \n",
    "    Returns:\n",
    "        (pd.Dataframe, MinMaxScaler):\n",
    "            The Data with `col` scaled, as well as the fit MinMaxScaler so that\n",
    "            data can be inverse transformed back into the original scale\n",
    "    \"\"\"\n",
    "    scaler = MinMaxScaler()\n",
    "    label_data = df[col].to_numpy().reshape(-1, 1)\n",
    "    scaler.fit(label_data)\n",
    "    df[col] = scaler.transform(label_data)\n",
    "    return df, scaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "dd10daa6-a385-44d5-a2e0-4b535b7c0a18",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>instant</th>\n",
       "      <th>dteday</th>\n",
       "      <th>season</th>\n",
       "      <th>yr</th>\n",
       "      <th>mnth</th>\n",
       "      <th>holiday</th>\n",
       "      <th>weekday</th>\n",
       "      <th>workingday</th>\n",
       "      <th>weathersit</th>\n",
       "      <th>temp</th>\n",
       "      <th>atemp</th>\n",
       "      <th>hum</th>\n",
       "      <th>windspeed</th>\n",
       "      <th>casual</th>\n",
       "      <th>registered</th>\n",
       "      <th>cnt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>2011-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0.344167</td>\n",
       "      <td>0.363625</td>\n",
       "      <td>0.805833</td>\n",
       "      <td>0.160446</td>\n",
       "      <td>331</td>\n",
       "      <td>654</td>\n",
       "      <td>985</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>2011-01-02</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0.363478</td>\n",
       "      <td>0.353739</td>\n",
       "      <td>0.696087</td>\n",
       "      <td>0.248539</td>\n",
       "      <td>131</td>\n",
       "      <td>670</td>\n",
       "      <td>801</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>2011-01-03</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.196364</td>\n",
       "      <td>0.189405</td>\n",
       "      <td>0.437273</td>\n",
       "      <td>0.248309</td>\n",
       "      <td>120</td>\n",
       "      <td>1229</td>\n",
       "      <td>1349</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   instant      dteday  season  yr  mnth  holiday  weekday  workingday  \\\n",
       "0        1  2011-01-01       1   0     1        0        6           0   \n",
       "1        2  2011-01-02       1   0     1        0        0           0   \n",
       "2        3  2011-01-03       1   0     1        0        1           1   \n",
       "\n",
       "   weathersit      temp     atemp       hum  windspeed  casual  registered  \\\n",
       "0           2  0.344167  0.363625  0.805833   0.160446     331         654   \n",
       "1           2  0.363478  0.353739  0.696087   0.248539     131         670   \n",
       "2           1  0.196364  0.189405  0.437273   0.248309     120        1229   \n",
       "\n",
       "    cnt  \n",
       "0   985  \n",
       "1   801  \n",
       "2  1349  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_file = \"data/day.csv\"\n",
    "daily_df = pd.read_csv(data_file)\n",
    "daily_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8897166-838f-4b20-ad95-1ea273ea97b8",
   "metadata": {},
   "source": [
    "### Visualization of the daily ridership counts over the span of the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "0276e574-6896-45fc-a8e6-69430c30d799",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 700x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(7, 4))\n",
    "ax.plot(daily_df[\"cnt\"])\n",
    "ax.set_xlabel(\"Day (Y-M-D)\")\n",
    "ax.set_ylabel(\"Rider count\")\n",
    "ax.set_title(\"Total Ridership per Day Through 2011 and 2012\")\n",
    "ax.set_xticks([0, len(daily_df)//2, len(daily_df)], labels=daily_df[\"dteday\"][[0, len(daily_df)//2, len(daily_df)-1]])\n",
    "fig.savefig(\"ridership.jpeg\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7dfe0fac-a85b-4b1a-80d9-ab21cb6c2742",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define which columns we want to keep for model training\n",
    "categorical_columns = [\"season\", \"holiday\", \"weekday\", \"workingday\", \"weathersit\"]\n",
    "continuous_columns = [\"temp\", \"atemp\", \"hum\", \"windspeed\"]\n",
    "feature_columns = categorical_columns + continuous_columns\n",
    "label_column = \"cnt\"\n",
    "# Only keep features and target\n",
    "daily_df = daily_df[feature_columns + [label_column]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0aeaca28-7083-4afa-986e-047e0430030f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>season_1</th>\n",
       "      <th>season_2</th>\n",
       "      <th>season_3</th>\n",
       "      <th>season_4</th>\n",
       "      <th>holiday_0</th>\n",
       "      <th>holiday_1</th>\n",
       "      <th>weekday_0</th>\n",
       "      <th>weekday_1</th>\n",
       "      <th>weekday_2</th>\n",
       "      <th>weekday_3</th>\n",
       "      <th>...</th>\n",
       "      <th>workingday_0</th>\n",
       "      <th>workingday_1</th>\n",
       "      <th>weathersit_1</th>\n",
       "      <th>weathersit_2</th>\n",
       "      <th>weathersit_3</th>\n",
       "      <th>temp</th>\n",
       "      <th>atemp</th>\n",
       "      <th>hum</th>\n",
       "      <th>windspeed</th>\n",
       "      <th>cnt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.344167</td>\n",
       "      <td>0.363625</td>\n",
       "      <td>0.805833</td>\n",
       "      <td>0.160446</td>\n",
       "      <td>0.110792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.363478</td>\n",
       "      <td>0.353739</td>\n",
       "      <td>0.696087</td>\n",
       "      <td>0.248539</td>\n",
       "      <td>0.089623</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.196364</td>\n",
       "      <td>0.189405</td>\n",
       "      <td>0.437273</td>\n",
       "      <td>0.248309</td>\n",
       "      <td>0.152669</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   season_1  season_2  season_3  season_4  holiday_0  holiday_1  weekday_0  \\\n",
       "0         1         0         0         0          1          0          0   \n",
       "1         1         0         0         0          1          0          1   \n",
       "2         1         0         0         0          1          0          0   \n",
       "\n",
       "   weekday_1  weekday_2  weekday_3  ...  workingday_0  workingday_1  \\\n",
       "0          0          0          0  ...             1             0   \n",
       "1          0          0          0  ...             1             0   \n",
       "2          1          0          0  ...             0             1   \n",
       "\n",
       "   weathersit_1  weathersit_2  weathersit_3      temp     atemp       hum  \\\n",
       "0             0             1             0  0.344167  0.363625  0.805833   \n",
       "1             0             1             0  0.363478  0.353739  0.696087   \n",
       "2             1             0             0  0.196364  0.189405  0.437273   \n",
       "\n",
       "   windspeed       cnt  \n",
       "0   0.160446  0.110792  \n",
       "1   0.248539  0.089623  \n",
       "2   0.248309  0.152669  \n",
       "\n",
       "[3 rows x 23 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Preprocess the data by one-hot encoding the categorical features\n",
    "# and MinMaxScale-ing the label, because it will also be used as a feature (the previous day's count)\n",
    "preprocessed_daily_df, daily_label_scaler = preprocess_bikeshare_data(daily_df, categorical_cols=categorical_columns, label_col=label_column)\n",
    "preprocessed_daily_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6b1f189f-60c4-4ad6-877f-71cd3f333b15",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "model_types = [\"baseline\", \"ridge\", \"feedforward\", \"lstm\"]\n",
    "BEST_MODEL_PERFORMANCES = {model_type:{\"train\": 0, \"test\": 0} for model_type in model_types}"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09d231f4-a8da-4323-9c95-6d29bec6578e",
   "metadata": {},
   "source": [
    "# 2. Non-LSTM Models Training and Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f1eace93-de5d-468a-9619-cd0e86714fc4",
   "metadata": {},
   "source": [
    "## 2.1 Create cross validation data for all models **except** LSTM model\n",
    "\n",
    "The LSTM model will be trained on hourly data instead of daily data to predict the next day's count. So, its data preparation will be a little different."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "80efd00c-6e0c-44da-bae6-85325e997079",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Inputs have shape: (730, 23)\n",
      "Labels have shape: (730, 1)\n"
     ]
    }
   ],
   "source": [
    "X = preprocessed_daily_df.iloc[:-1, :].to_numpy()\n",
    "y = preprocessed_daily_df[[label_column]].iloc[1:, :].to_numpy()\n",
    "\n",
    "print(f\"Inputs have shape: {X.shape}\")\n",
    "print(f\"Labels have shape: {y.shape}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "087eceeb-e35c-422f-8d7b-2e1236a32bc2",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def create_timeseries_crossval_splits(data, n_splits):\n",
    "    \"\"\"Create train and validation sets for timeseries cross validation\n",
    "    \n",
    "    Args:\n",
    "        data (ndarray): 2D array of input samples\n",
    "        n_splits (int): Number of subsets for cross validation\n",
    "    \n",
    "    Returns:\n",
    "        list[tuple]: Each tuple is (TRAIN_INDICES, VALIDATION_INDICES)\n",
    "    \"\"\"\n",
    "    timeseries_crossval = TimeSeriesSplit(n_splits=5)\n",
    "    return list(timeseries_crossval.split(data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ff20ac01-adc6-4cf7-8f07-fa0435e99e0f",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total dataset size: 730 samples\n",
      "Train and validation set size: 630 samples\n",
      "Holdout test size: 100 samples\n"
     ]
    }
   ],
   "source": [
    "test_set_size = 100  # Last 100 days of data\n",
    "train_set_size = len(X) - test_set_size\n",
    "\n",
    "X_train_and_val = X[:train_set_size]\n",
    "y_train_and_val = y[:train_set_size]\n",
    "X_test = X[train_set_size:]\n",
    "y_test = y[train_set_size:]\n",
    "\n",
    "# Create TRAIN, VALIDATION, and TEST sets for time series cross validation\n",
    "crossval_indices = create_timeseries_crossval_splits(X_train_and_val, n_splits=5)\n",
    "\n",
    "print(f\"Total dataset size: {len(X)} samples\")\n",
    "print(f\"Train and validation set size: {len(X_train_and_val)} samples\")\n",
    "print(f\"Holdout test size: {len(X_test)} samples\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9612e448-c6e1-4d93-99f3-1225c5e6a0ec",
   "metadata": {},
   "source": [
    "## 2.2 Naive Baseline Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "37e86857-f82d-4fb4-a5f6-21c7aeb35bd5",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def baseline_model(input_vectors, cnt_col):\n",
    "    \"\"\"A naive model which simply uses the previous day's count\n",
    "       as the prediction for today\"\"\"\n",
    "    return input_vectors[:, cnt_col]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "03315818-e2bb-417e-9fa5-7d5990eb9444",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def get_rmse_baseline_model(model, X, y, scaler, cnt_col):\n",
    "    baseline_predictions = model(X, cnt_col)\n",
    "    rmse = np.average(mean_squared_error(y, baseline_predictions, squared=False))\n",
    "    # The count was MinMaxScaled to use in model training. But for better interpretability of\n",
    "    # the results, it is transformed back to its original value here\n",
    "    rmse = scaler.inverse_transform(rmse.reshape(-1, 1))\n",
    "    return rmse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e4c3afb3-c51a-4726-b0a5-0f1b59a14211",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1035.6402824583395\n",
      "1363.4416200491169\n"
     ]
    }
   ],
   "source": [
    "# The column with the previous day's count is the last item in the vector\n",
    "RIDERSHIP_COUNT_IDX = X.shape[1] - 1  \n",
    "\n",
    "baseline_rmse_train = get_rmse_baseline_model(baseline_model, X_train_and_val, y_train_and_val, daily_label_scaler, RIDERSHIP_COUNT_IDX)[0][0]\n",
    "baseline_rmse_test = get_rmse_baseline_model(baseline_model, X_test, y_test, daily_label_scaler, RIDERSHIP_COUNT_IDX)[0][0]\n",
    "\n",
    "print(baseline_rmse_train)\n",
    "print(baseline_rmse_test)\n",
    "\n",
    "BEST_MODEL_PERFORMANCES[\"baseline\"][\"train\"] = baseline_rmse_train\n",
    "BEST_MODEL_PERFORMANCES[\"baseline\"][\"test\"] = baseline_rmse_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "8471aef2-a22c-41d1-854c-046ad4809521",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'baseline': {'train': 1035.6402824583395, 'test': 1363.4416200491169},\n",
       " 'ridge': {'train': 0, 'test': 0},\n",
       " 'feedforward': {'train': 0, 'test': 0},\n",
       " 'lstm': {'train': 0, 'test': 0}}"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "BEST_MODEL_PERFORMANCES"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4923885f-3c99-4379-b210-0e04bc1c637c",
   "metadata": {},
   "source": [
    "## 2.3 Ridge Regression Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "bbe11d68-0021-4e5e-a487-f7277487398c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def evaluate_linear_model(model, X, y, ts_crossvalidation_indices):\n",
    "    \"\"\"Evaluate a linear model using timeseries cross validation\n",
    "    \n",
    "    Args:\n",
    "        model: A class which can be trained with .fit() and has .predict() to make predictions\n",
    "        X (ndarray): A 2D array of float values, inputs to the model\n",
    "        y (ndarray): A 1d array of float values, labels for the inputs\n",
    "        ts_crossvalidation_indices(list[tuple]): A list of indices for training and testing sets\n",
    "                                                 to perform timeseries cross validation\n",
    "    \n",
    "    Returns:\n",
    "        list[float], model: RMSE scores for each cross validation subset, and the final trained model\n",
    "    \"\"\"\n",
    "    rmse_scores = []\n",
    "    for train_indices, validation_indices in ts_crossvalidation_indices:\n",
    "        model = sklearn.base.clone(model)  # Reset the model parameters\n",
    "        X_train, y_train = X[train_indices], y[train_indices]\n",
    "        X_validation, y_validation = X[validation_indices], y[validation_indices]\n",
    "        model.fit(X_train, y_train)\n",
    "        predictions = model.predict(X_validation)\n",
    "        rmse_scores.append(mean_squared_error(y_validation, predictions, squared=False))\n",
    "    return rmse_scores, model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "7bdff461-e202-492f-be39-24f6c09ec8b3",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "alphas = [1e-5, 1e-4, 1e-3, 1e-2, 1e-1, 1, 10]\n",
    "avg_rmse_by_alpha = []\n",
    "ridge_models = []\n",
    "for alpha in alphas:\n",
    "    rmse_scores_ridge, model = evaluate_linear_model(linear_model.Ridge(alpha=alpha), X_train_and_val, y_train_and_val, crossval_indices)\n",
    "    avg_rmse_ridge = daily_label_scaler.inverse_transform(np.average(rmse_scores_ridge).reshape(-1, 1))\n",
    "    avg_rmse_by_alpha.append(avg_rmse_ridge[0][0])\n",
    "    ridge_models.append(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "bfdff5c0-6576-4966-8f5c-80dfe7e418cd",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.plot(avg_rmse_by_alpha, label=\"Average RMSE\")\n",
    "ax.set_xticks(range(len(alphas)))\n",
    "ax.set_xticklabels(alphas)\n",
    "ax.set_xlabel(\"alpha\")\n",
    "ax.set_ylabel(\"Mean RMSE\")\n",
    "ax.set_title(\"Ridge Regression Mean Cross Validation RMSE by alpha\")\n",
    "ax.legend()\n",
    "fig.savefig(\"ridge_regression_hyperparameter_tuning.jpeg\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d94cebf-8c02-4292-9203-909932e587aa",
   "metadata": {},
   "source": [
    "### 2.3.1 Evaluate Best Ridge Regression Model on Train and Test Sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e149cb8a-be6f-4d51-b79b-10f3ed08bca5",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ridge(alpha=0.01)\n"
     ]
    }
   ],
   "source": [
    "best_model_index = np.argmin(avg_rmse_by_alpha)\n",
    "best_ridge_model = ridge_models[best_model_index]\n",
    "print(best_ridge_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "490e72fd-42ee-473b-876b-d4e914fbec24",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def get_rmse_ridge_model(model, X, y, scaler):\n",
    "    \"\"\"Compute RMSE for the ridge model\"\"\"\n",
    "    ridge_predictions = model.predict(X)\n",
    "    rmse = np.average(mean_squared_error(y, ridge_predictions, squared=False))\n",
    "    # The count was MinMaxScaled to use in model training. But for better interpretability of\n",
    "    # the results, it is transformed back to its original value here\n",
    "    rmse = daily_label_scaler.inverse_transform(rmse.reshape(-1, 1))\n",
    "    return rmse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "6fe5d619-df33-4d40-9281-a19e60f52103",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "969.0154216038247\n",
      "1307.4153172253148\n"
     ]
    }
   ],
   "source": [
    "best_model_rmse_train = get_rmse_ridge_model(best_ridge_model, X_train_and_val, y_train_and_val, daily_label_scaler)[0][0]\n",
    "best_model_rmse_test = get_rmse_ridge_model(best_ridge_model, X_test, y_test, daily_label_scaler)[0][0]\n",
    "print(best_model_rmse_train)\n",
    "print(best_model_rmse_test)\n",
    "\n",
    "BEST_MODEL_PERFORMANCES[\"ridge\"][\"train\"] = best_model_rmse_train\n",
    "BEST_MODEL_PERFORMANCES[\"ridge\"][\"test\"] = best_model_rmse_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "bf1aa184-825c-4740-b864-e150231eb893",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'baseline': {'train': 1035.6402824583395, 'test': 1363.4416200491169},\n",
       " 'ridge': {'train': 969.0154216038247, 'test': 1307.4153172253148},\n",
       " 'feedforward': {'train': 0, 'test': 0},\n",
       " 'lstm': {'train': 0, 'test': 0}}"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "BEST_MODEL_PERFORMANCES"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "358259c2-6f54-47d7-8b46-6bd764ccfcea",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 2.4 Feedforward Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "ef913afc-07b2-4516-8367-e4f2d0349133",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch import nn\n",
    "from torch.utils.data import Dataset\n",
    "from torch.utils.data import DataLoader"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed9b7b32-3bfc-46a4-96dd-3aa37c7c55b1",
   "metadata": {},
   "source": [
    "### 2.4.1 Model Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "746412f4-369c-46ca-a33c-191d116cf122",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "class BikeshareDataset(Dataset):\n",
    "    def __init__(self, X, y):\n",
    "        self.X = torch.from_numpy(X).float()\n",
    "        self.y = torch.from_numpy(y).float()\n",
    "        self.dataset_size = len(self.y)\n",
    "    \n",
    "    def __len__(self):\n",
    "        return self.dataset_size\n",
    "        \n",
    "    def __getitem__(self, idx):\n",
    "        feature_vec = self.X[idx]\n",
    "        label = self.y[idx]\n",
    "        return feature_vec, label\n",
    "    \n",
    "    \n",
    "class NeuralNetwork(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(NeuralNetwork, self).__init__()\n",
    "        self.linear_relu_stack = nn.Sequential(\n",
    "            nn.Linear(23, 32),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(32, 32),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(32, 1)\n",
    "        )\n",
    "\n",
    "    def forward(self, x):\n",
    "        return self.linear_relu_stack(x)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "f6f338d6-a604-4e91-908c-e191457fd78f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def train_loop(dataloader, model, loss_fn, optimizer):\n",
    "    \"\"\"Train the neural net\"\"\"\n",
    "    size = len(dataloader.dataset)\n",
    "    for batch, (X, y) in enumerate(dataloader):\n",
    "        pred = model(X)\n",
    "        loss = loss_fn(pred, y)\n",
    "\n",
    "        # Backpropagation\n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "\n",
    "def test_loop(dataloader, model):\n",
    "    \"\"\"Evaluate the neural net on a holdout set\"\"\"\n",
    "    size = len(dataloader.dataset)\n",
    "    num_batches = len(dataloader)\n",
    "    error = 0\n",
    "    with torch.no_grad():\n",
    "        for X, y in dataloader:\n",
    "            pred = model(X).numpy()\n",
    "            error += mean_squared_error(y.numpy(), pred, squared=False)   \n",
    "    error /= num_batches\n",
    "    return error"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "530e1c44-f40c-4ebe-a010-f6aa5e9660fe",
   "metadata": {},
   "source": [
    "### 2.4.2 Run Cross Validation With Different Hyperparameter Configurations to Find Best Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "374f8c40-4e9d-4d4b-95c6-a9fc0572e705",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training with configuration: (loss_fn=MSELoss(), lr=0.01)\n",
      "Training with configuration: (loss_fn=MSELoss(), lr=0.001)\n",
      "Training with configuration: (loss_fn=L1Loss(), lr=0.01)\n",
      "Training with configuration: (loss_fn=L1Loss(), lr=0.001)\n"
     ]
    }
   ],
   "source": [
    "n_epochs = 50\n",
    "configurations = [(nn.MSELoss(), 1e-2),\n",
    "                  (nn.MSELoss(), 1e-3),\n",
    "                  (nn.L1Loss(), 1e-2),\n",
    "                  (nn.L1Loss(), 1e-3)]\n",
    "\n",
    "mean_validation_rmse_by_configuration = []\n",
    "\n",
    "# Run cross validation for each hyperparameter configuration.\n",
    "# For each configuration, train K models over all of the cross validation splits\n",
    "# and compute the average RMSE. The average RMSE over the cross validation splits\n",
    "# is that configuration's score.\n",
    "for loss_fn, learning_rate in configurations:\n",
    "    print(f\"Training with configuration: (loss_fn={loss_fn}, lr={learning_rate})\")\n",
    "    rmse_scores_feedforward_nn = []\n",
    "    for train_indices, validation_indices in crossval_indices:\n",
    "        model = NeuralNetwork()  # Create a new model to train\n",
    "        optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)\n",
    "        train_dataloader = DataLoader(BikeshareDataset(X_train_and_val[train_indices], y_train_and_val[train_indices]), batch_size=64, shuffle=False)\n",
    "        validation_dataloader = DataLoader(BikeshareDataset(X_train_and_val[validation_indices], y_train_and_val[validation_indices]), batch_size=64, shuffle=False)\n",
    "        epochs = n_epochs\n",
    "        for _ in range(epochs):\n",
    "            train_loop(train_dataloader, model, loss_fn, optimizer)\n",
    "        rmse_scores_feedforward_nn.append(test_loop(validation_dataloader, model))\n",
    "\n",
    "    mean_rmse_feedforward_nn = daily_label_scaler.inverse_transform(np.average(rmse_scores_feedforward_nn).reshape(-1, 1))\n",
    "    mean_validation_rmse_by_configuration.append((mean_rmse_feedforward_nn[0][0], (loss_fn, learning_rate), model))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "bc871562-b8ab-450b-ac04-703ec521c26f",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(L1Loss(), 0.01)\n"
     ]
    }
   ],
   "source": [
    "best_rmse, best_config, best_model = min(mean_validation_rmse_by_configuration, key = lambda t: t[0])\n",
    "print(best_config)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b844d24c-ea24-4110-aea4-40ecc41d56ff",
   "metadata": {},
   "source": [
    "### 2.4.3 Evaluate the Best Model on Train and Test Sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "77be801c-4799-474e-8cee-76154abcbe7b",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def get_rmse_feedforward_nn_model(model, dataloader, scaler):\n",
    "    rmse = test_loop(dataloader, model)\n",
    "    rmse = daily_label_scaler.inverse_transform(rmse.reshape(-1, 1))\n",
    "    return rmse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "39b4051a-7202-49f8-9aa1-739bc85c15ae",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1387.8865265667437\n",
      "1764.8685614764688\n"
     ]
    }
   ],
   "source": [
    "train_dataloader = DataLoader(BikeshareDataset(X_train_and_val, y_train_and_val), batch_size=64, shuffle=False)\n",
    "test_dataloader = DataLoader(BikeshareDataset(X_test, y_test), batch_size=64, shuffle=False)\n",
    "\n",
    "best_model_rmse_train = get_rmse_feedforward_nn_model(best_model, train_dataloader, daily_label_scaler)[0][0]\n",
    "best_model_rmse_test = get_rmse_feedforward_nn_model(best_model, test_dataloader, daily_label_scaler)[0][0]\n",
    "print(best_model_rmse_train)\n",
    "print(best_model_rmse_test)\n",
    "\n",
    "BEST_MODEL_PERFORMANCES[\"feedforward\"][\"train\"] = best_model_rmse_train\n",
    "BEST_MODEL_PERFORMANCES[\"feedforward\"][\"test\"] = best_model_rmse_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "54395725-7596-49ca-b591-983c99af1210",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'baseline': {'train': 1035.6402824583395, 'test': 1363.4416200491169},\n",
       " 'ridge': {'train': 969.0154216038247, 'test': 1307.4153172253148},\n",
       " 'feedforward': {'train': 1387.8865265667437, 'test': 1764.8685614764688},\n",
       " 'lstm': {'train': 0, 'test': 0}}"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "BEST_MODEL_PERFORMANCES"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "97064358-1d1d-42bf-aa7c-3b39af2bf6fd",
   "metadata": {},
   "source": [
    "# 3. Long Short Term Memory (LSTM) Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "a13026ef-4f8b-47b4-9c4d-66d8842474ae",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "os.environ[\"TF_CPP_MIN_LOG_LEVEL\"] = \"3\"  # Suppress tensorflow debugging info\n",
    "import keras\n",
    "from keras.utils import timeseries_dataset_from_array\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, LSTM, Dropout\n",
    "from keras.optimizers import Adam, SGD"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "60789f17-b555-4667-bda1-2027d9bfdae6",
   "metadata": {},
   "source": [
    "### 3.1 Data Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "c74b091c-780d-4c21-be4a-eead8cf11228",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "data_file = \"data/hour.csv\"\n",
    "hourly_df = pd.read_csv(data_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "86805e64-a274-41f2-9466-be057581ca1c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def nearest_fill_missing_hours(data):\n",
    "    \"\"\"Fill missing hours with the nearest available data\"\"\"\n",
    "    data_copy = data.copy()\n",
    "    data_copy = (data_copy\n",
    "                 .set_index(\"hr\")\n",
    "                 .reindex(list(range(24)), method=\"nearest\")\n",
    "                 .reset_index())\n",
    "    return data_copy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "f280a4c4-9d20-4d87-a572-d51ec1aacf52",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>hr</th>\n",
       "      <th>instant</th>\n",
       "      <th>dteday</th>\n",
       "      <th>season</th>\n",
       "      <th>yr</th>\n",
       "      <th>mnth</th>\n",
       "      <th>holiday</th>\n",
       "      <th>weekday</th>\n",
       "      <th>workingday</th>\n",
       "      <th>weathersit</th>\n",
       "      <th>temp</th>\n",
       "      <th>atemp</th>\n",
       "      <th>hum</th>\n",
       "      <th>windspeed</th>\n",
       "      <th>casual</th>\n",
       "      <th>registered</th>\n",
       "      <th>cnt</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2011-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.2879</td>\n",
       "      <td>0.81</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3</td>\n",
       "      <td>13</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2011-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.22</td>\n",
       "      <td>0.2727</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.0</td>\n",
       "      <td>8</td>\n",
       "      <td>32</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2011-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.22</td>\n",
       "      <td>0.2727</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5</td>\n",
       "      <td>27</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>2011-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.2879</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>2011-01-01</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.24</td>\n",
       "      <td>0.2879</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   hr  instant      dteday  season  yr  mnth  holiday  weekday  workingday  \\\n",
       "0   0        1  2011-01-01       1   0     1        0        6           0   \n",
       "1   1        2  2011-01-01       1   0     1        0        6           0   \n",
       "2   2        3  2011-01-01       1   0     1        0        6           0   \n",
       "3   3        4  2011-01-01       1   0     1        0        6           0   \n",
       "4   4        5  2011-01-01       1   0     1        0        6           0   \n",
       "\n",
       "   weathersit  temp   atemp   hum  windspeed  casual  registered  cnt  \n",
       "0           1  0.24  0.2879  0.81        0.0       3          13   16  \n",
       "1           1  0.22  0.2727  0.80        0.0       8          32   40  \n",
       "2           1  0.22  0.2727  0.80        0.0       5          27   32  \n",
       "3           1  0.24  0.2879  0.75        0.0       3          10   13  \n",
       "4           1  0.24  0.2879  0.75        0.0       0           1    1  "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The hourly data has missing values, so fill them with the nearest available data\n",
    "hourly_df = hourly_df.groupby(\"dteday\").apply(nearest_fill_missing_hours).reset_index(drop=True)\n",
    "hourly_df.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "92cd3f7c-3cfc-4c43-ba35-a6e2b3b6bde5",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# We'll need the total count on each day, it is the label\n",
    "count_by_day = hourly_df[[\"dteday\", \"cnt\"]].groupby(\"dteday\").sum()\n",
    "\n",
    "# Remove any unneeded columns\n",
    "hourly_df = hourly_df[feature_columns + [label_column]]\n",
    "preprocessed_hourly_df, hourly_label_scaler = preprocess_bikeshare_data(hourly_df, categorical_cols=categorical_columns, label_col=label_column)\n",
    "\n",
    "# Apply the MinMaxScaler to the daily counts\n",
    "count_by_day[\"cnt\"] = hourly_label_scaler.transform(np.array(count_by_day[\"cnt\"]).reshape(-1, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "1050c3b6-e085-43dd-a2d5-f2161e505300",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "season         1.0000\n",
       "holiday        0.0000\n",
       "weekday        6.0000\n",
       "workingday     0.0000\n",
       "weathersit     1.0000\n",
       "temp           0.2400\n",
       "atemp          0.2879\n",
       "hum            0.8100\n",
       "windspeed      0.0000\n",
       "cnt           16.0000\n",
       "Name: 0, dtype: float64"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hourly_df.iloc[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8894a2b7-cdda-4a35-bfc4-94331885a1ca",
   "metadata": {},
   "source": [
    "### 3.2 Split into Training, Validation, and Test Sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "50c8f081-c3f8-4366-be5d-e47fa57ed2b2",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X shape: (17520, 24)\n",
      "y shape: (730,)\n"
     ]
    }
   ],
   "source": [
    "HOURS_PER_DAY = 24\n",
    "TOTAL_NUM_DAYS = len(hourly_df) / HOURS_PER_DAY\n",
    "\n",
    "# Input samples will be all except the hourly measurements taken on the last day,\n",
    "# because we don't have a label for that day (i.e., no data for the day after it)\n",
    "X_lstm = preprocessed_hourly_df.iloc[:-HOURS_PER_DAY, :].to_numpy()\n",
    "\n",
    "# Labels will be the sum of each 24 hour period. I.e., total ridership in one day.\n",
    "# CANNOT use the raw daily data for this, as hourly values have been nearest-filled\n",
    "y_lstm = np.apply_along_axis(sum, 1, np.array_split(np.array(preprocessed_hourly_df[\"cnt\"][HOURS_PER_DAY:]), TOTAL_NUM_DAYS - 1))\n",
    "\n",
    "print(f\"X shape: {X_lstm.shape}\")\n",
    "print(f\"y shape: {y_lstm.shape}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "e21bde81-2413-4d86-b326-b0440403e55d",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# The test set will be the last 100 days of data\n",
    "lstm_test_set_size = HOURS_PER_DAY * 100\n",
    "lstm_train_set_size = len(X_lstm) - lstm_test_set_size\n",
    "\n",
    "X_train_and_val_lstm = X_lstm[:lstm_train_set_size]\n",
    "y_train_and_val_lstm = y_lstm[:(lstm_train_set_size // HOURS_PER_DAY)]\n",
    "X_test_lstm = X_lstm[lstm_train_set_size:]\n",
    "y_test_lstm = y_lstm[(lstm_train_set_size // HOURS_PER_DAY):]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "cdfb7371-25a7-439d-bb2c-e1be26dc3e26",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "crossval_indices = create_timeseries_crossval_splits(X_train_and_val_lstm, n_splits=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "e9f7e178-d445-4ab3-9be7-4d0690d9d4a9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Metal device set to: Apple M1 Pro\n",
      "There are 5 splits for LSTM timeseries cross validation\n"
     ]
    }
   ],
   "source": [
    "BATCH_SIZE=64\n",
    "\n",
    "lstm_datasets = []\n",
    "for train_indices, validation_indices in crossval_indices:\n",
    "    X_train = X_train_and_val_lstm[train_indices]\n",
    "    y_train = y_train_and_val_lstm[train_indices // HOURS_PER_DAY]\n",
    "    X_val = X_train_and_val_lstm[validation_indices]\n",
    "    y_val = y_train_and_val_lstm[validation_indices // HOURS_PER_DAY]\n",
    "    \n",
    "    train_dataset = timeseries_dataset_from_array(\n",
    "        data=X_train,\n",
    "        targets=y_train,\n",
    "        sequence_length=HOURS_PER_DAY,\n",
    "        batch_size=BATCH_SIZE)\n",
    "\n",
    "    validation_dataset = timeseries_dataset_from_array(\n",
    "        data=X_val,\n",
    "        targets=y_val,\n",
    "        sequence_length=HOURS_PER_DAY,\n",
    "        batch_size=BATCH_SIZE)\n",
    "\n",
    "    lstm_datasets.append((train_dataset, validation_dataset))\n",
    "    \n",
    "print(f\"There are {len(lstm_datasets)} splits for LSTM timeseries cross validation\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "000a407f-c1cd-4f8b-904b-8134805db417",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def build_lstm_model(input_shape, nodes_per_layer, lr):\n",
    "    \"\"\"Helper to enable trying different hyperparameters to find best configuration\n",
    "    \n",
    "    Args:\n",
    "        input_shape (tuple): Shape of the input tensors\n",
    "        nodes_per_layer (int): Number of nodes in each of the 2 hidden layers\n",
    "        lr (float): Learning rate\n",
    "    \n",
    "    Returns:\n",
    "        Sequential: A Keras Sequential LSTM model\n",
    "    \"\"\"\n",
    "    model = Sequential()\n",
    "    model.add(keras.Input(shape=input_shape))  # `None` indicates the sequence is of arbitrary length\n",
    "    model.add(LSTM(nodes_per_layer, return_sequences=True))\n",
    "    model.add(Dropout(0.15))\n",
    "    model.add(LSTM(nodes_per_layer))\n",
    "    model.add(Dropout(0.15))\n",
    "    model.add(Dense(1))\n",
    "    model.compile(optimizer=SGD(learning_rate=lr),\n",
    "                  loss=\"mean_squared_error\", \n",
    "                  metrics=[\"mse\"])\n",
    "    model.build()\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "35b60602-6791-4b88-9583-a46b6ede0982",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "LSTM_CONFIGURATIONS = [(16, 1e-3),\n",
    "                       (16, 1e-4),\n",
    "                       (32, 1e-3),\n",
    "                       (32, 1e-4)]\n",
    "\n",
    "FEATURE_VEC_SIZE = 24\n",
    "model = build_lstm_model(input_shape=(HOURS_PER_DAY, FEATURE_VEC_SIZE), nodes_per_layer=16, lr=1e-2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "5ee9c136-8bf1-4302-85c4-67bd60bb3542",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running cross validation with configuration: nodes=16, lr=0.001\n",
      "Running cross validation with configuration: nodes=16, lr=0.0001\n",
      "Running cross validation with configuration: nodes=32, lr=0.001\n",
      "Running cross validation with configuration: nodes=32, lr=0.0001\n",
      "CPU times: user 8min 57s, sys: 4min 29s, total: 13min 27s\n",
      "Wall time: 5min 57s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "configs_histories_rmse_models = []\n",
    "for nodes_per_layer, learning_rate in LSTM_CONFIGURATIONS:\n",
    "    print(f\"Running cross validation with configuration: nodes={nodes_per_layer}, lr={learning_rate}\")\n",
    "    histories = []\n",
    "    for i, (train_dataset, validation_dataset) in enumerate(lstm_datasets):\n",
    "        history = model.fit(train_dataset, epochs=10, validation_data=validation_dataset, verbose=0)\n",
    "        histories.append(history.history)\n",
    "    root_rmses = np.sqrt([hist[\"val_mse\"] for hist in histories])\n",
    "    avg_rmse = np.average(root_rmses)\n",
    "    configs_histories_rmse_models.append(((nodes_per_layer, learning_rate), histories[-1], avg_rmse, model))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "d0e35497-e386-407f-9dff-3a7893ed59a2",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "best_config, best_history, best_rmse, best_lstm_model = min(configs_histories_rmse_models, key = lambda t: t[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "20ba957f-1a2b-47ed-a0e4-c4c4f755b956",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(32, 0.001)\n"
     ]
    }
   ],
   "source": [
    "print(best_config)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f22ce1c8-699e-4d42-8333-2bffb9fea000",
   "metadata": {},
   "source": [
    "### 3.3 Evaluate the Best LSTM Model Configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "752a4117-d9a2-42b2-b78f-59ba172c3e20",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "eval_train_dataset = timeseries_dataset_from_array(\n",
    "    data=X_train_and_val_lstm[train_indices],\n",
    "    targets=y_train_and_val_lstm[train_indices // HOURS_PER_DAY],\n",
    "    sequence_length=HOURS_PER_DAY,\n",
    "    batch_size=BATCH_SIZE)\n",
    "\n",
    "eval_test_dataset = timeseries_dataset_from_array(\n",
    "    data=X_test_lstm,\n",
    "    targets=y_test_lstm,\n",
    "    sequence_length=HOURS_PER_DAY,\n",
    "    batch_size=BATCH_SIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "45b6928d-0afe-4562-8861-406a13a115ca",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def evaluate_lstm_model(dataset, model, scaler):\n",
    "    batch_errors = []\n",
    "    for X, y in dataset:\n",
    "        predictions = model(X).numpy()\n",
    "        rmse = np.average(mean_squared_error(y.numpy(), predictions, squared=False))\n",
    "        rmse = scaler.inverse_transform(rmse.reshape(-1, 1))\n",
    "        batch_errors.append(rmse)\n",
    "    return np.average(batch_errors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "c8e338f6-9ed9-430e-a2e9-904ed00f703a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1879.1334771233016\n",
      "2090.9435168808855\n"
     ]
    }
   ],
   "source": [
    "best_lstm_model_rmse_train = evaluate_lstm_model(eval_train_dataset, best_lstm_model, hourly_label_scaler)\n",
    "best_lstm_model_rmse_test = evaluate_lstm_model(eval_test_dataset, best_lstm_model, hourly_label_scaler)\n",
    "\n",
    "print(best_lstm_model_rmse_train)\n",
    "print(best_lstm_model_rmse_test)\n",
    "\n",
    "BEST_MODEL_PERFORMANCES[\"lstm\"][\"train\"] = best_lstm_model_rmse_train\n",
    "BEST_MODEL_PERFORMANCES[\"lstm\"][\"test\"] = best_lstm_model_rmse_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "e2201572-aa4e-481f-95c2-2f8a3ac8e5d9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'baseline': {'train': 1035.6402824583395, 'test': 1363.4416200491169},\n",
       " 'ridge': {'train': 969.0154216038247, 'test': 1307.4153172253148},\n",
       " 'feedforward': {'train': 1387.8865265667437, 'test': 1764.8685614764688},\n",
       " 'lstm': {'train': 1879.1334771233016, 'test': 2090.9435168808855}}"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "BEST_MODEL_PERFORMANCES"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7e8fd57-1961-46d1-8ba4-cb8d0ae34a73",
   "metadata": {},
   "source": [
    "# 4. Compare Performance of Best Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "08f1e703-00ea-4635-8be3-890ddf87c5aa",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_names = [\"baseline\", \"ridge\", \"feedforward\", \"lstm\"]\n",
    "all_train_rmses = [np.around(BEST_MODEL_PERFORMANCES[model_type][\"train\"]) for model_type in model_names]\n",
    "all_test_rmses = [np.around(BEST_MODEL_PERFORMANCES[model_type][\"test\"]) for model_type in model_names]\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "train_bars = ax.bar([0.9, 1.9, 2.9, 3.9], all_train_rmses, width=0.2, label=\"Train Set\")\n",
    "ax.bar_label(train_bars)\n",
    "test_bars = ax.bar([1.1, 2.1, 3.1, 4.1], all_test_rmses, width=0.2, label=\"Test Set\")\n",
    "ax.bar_label(test_bars)\n",
    "ax.set_title(\"Root Mean Squared Error of Best Models (Lower is Better)\")\n",
    "ax.set_xlabel(\"Model\")\n",
    "ax.set_ylabel(\"RMSE of Best Configuration\")\n",
    "ax.set_xticks([1, 2, 3, 4])\n",
    "ax.set_xticklabels([\"Baseline\", \"Ridge Regression\", \"Feedforward NN\", \"LSTM\"])\n",
    "ax.legend()\n",
    "fig.savefig(\"best_model_performances.jpeg\")\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3e133d99-9d09-420d-b2a1-b58e760a7c70",
   "metadata": {},
   "source": [
    "# Discussion\n",
    "\n",
    "Perhaps counterintuitively, the more sophisticated neural network models performed significantly worse than the simple ridge regression model. Even the naive baseline \"model\" outperformed them.\n",
    "\n",
    "The simplest explanation in this case is that the training set was too small for the neural networks to generalize well. However, even with a larger dataset there simply may not be a strong temporal relationship in the data. Having samples from previous hours and days may help capture cyclical or seasonal influences, such as the general trend of higher ridership in warmer months and lower ridership in colder months, but it is likely not helpful with respect to other features. Take the weather, for example. Does the fact it was raining yesterday have a strong influence on ridership today? I doubt it, even if the ground is still a little wet. Similarly, if it is raining today it does not matter that the weather was clear yesterday.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c755f717-bdd3-489d-aac5-aec6f944186c",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "bikeshare_modelling",
   "language": "python",
   "name": "bikeshare_modelling"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}