STG-NCDE-main/STG-NCDE - Historical average baselines-FINAL-seq2seq.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/rodr/code\n",
      "Namespace(batch_size=64, cheb_k=2, column_wise=False, comment='', cuda=True, dataset='PEMSD4', debug=False, default_graph=True, device=0, early_stop=True, early_stop_patience=15, embed_dim=10, epochs=100, f='/home/rodr/.local/share/jupyter/runtime/kernel-8ff59059-8e28-48ee-83ec-fce23fc5d639.json', g_type='agc', grad_norm=False, hid_dim=128, hid_hid_dim=128, horizon=12, input_dim=2, lag=12, log_dir='../runs', log_step=20, loss_func='mae', lr_decay=False, lr_decay_rate=0.3, lr_decay_step='5,20,40,70', lr_init=0.001, mae_thresh=None, mape_thresh=0.0, max_grad_norm=5, missing_rate=0.1, missing_test=False, mode='train', model='GCDE', model_path='', model_type='type1', normalizer='std', num_layers=3, num_nodes=307, output_dim=1, plot=False, real_value=True, seed=10, solver='rk4', teacher_forcing=False, tensorboard=False, test_ratio=0.2, tod=False, val_ratio=0.2, weight_decay=0.001)\n",
      "Load PEMSD4 Dataset shaped:  (16992, 307, 1) 919.0 0.0 211.7007794815878 180.0\n",
      "(16992, 307, 1)\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import sys\n",
    "file_dir = os.path.dirname(os.path.dirname(os.path.abspath(\"/home/rodr/code/mobility-baselines/STG-NCDE-main\")))\n",
    "print(file_dir)\n",
    "sys.path.append(file_dir)\n",
    "\n",
    "import torch\n",
    "import numpy as np\n",
    "import torch.nn as nn\n",
    "import argparse\n",
    "import configparser\n",
    "import time\n",
    "from matplotlib import pyplot as plt\n",
    "import datetime\n",
    "\n",
    "from sklearn.linear_model import LinearRegression, Lasso\n",
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "\n",
    "from model.BasicTrainer_cde import Trainer\n",
    "from lib.TrainInits import init_seed\n",
    "from lib.dataloader import *\n",
    "from lib.TrainInits import print_model_parameters\n",
    "import os\n",
    "from os.path import join\n",
    "from model.Make_model import make_model\n",
    "from torch.utils.tensorboard import SummaryWriter\n",
    "#*************************************************************************#\n",
    "Mode = 'train'\n",
    "DEBUG = 'False'\n",
    "DATASET = 'PEMSD4'      #PEMSD4 or PEMSD8\n",
    "MODEL = 'GCDE'\n",
    "\n",
    "#get configuration\n",
    "config_file = 'model/{}_{}.conf'.format(DATASET, MODEL)\n",
    "#print('Read configuration file: %s' % (config_file))\n",
    "config = configparser.ConfigParser()\n",
    "config.read(config_file)\n",
    "\n",
    "from lib.metrics import MAE_torch\n",
    "def masked_mae_loss(scaler, mask_value):\n",
    "    def loss(preds, labels):\n",
    "        if scaler:\n",
    "            preds = scaler.inverse_transform(preds)\n",
    "            labels = scaler.inverse_transform(labels)\n",
    "        mae = MAE_torch(pred=preds, true=labels, mask_value=mask_value)\n",
    "        return mae\n",
    "    return loss\n",
    "\n",
    "# python Run_cde.py --dataset='PEMSD4' --model='GCDE' --model_type='type1' --embed_dim=10 --hid_dim=64 --hid_hid_dim=64 --num_layers=2 --lr_init=0.001 --weight_decay=1e-3 --epochs=200 --tensorboard --comment=\"\" --device=0\n",
    "\n",
    "#parser\n",
    "args = argparse.ArgumentParser(description='arguments')\n",
    "args.add_argument('--dataset', default=DATASET, type=str)\n",
    "args.add_argument('--mode', default=Mode, type=str)\n",
    "args.add_argument('--device', default=0, type=int, help='indices of GPUs')\n",
    "args.add_argument('--debug', default=DEBUG, type=eval)\n",
    "args.add_argument('--model', default=MODEL, type=str)\n",
    "args.add_argument('--cuda', default=True, type=bool)\n",
    "args.add_argument('--comment', default='', type=str)\n",
    "args.add_argument(\"-f\", type=str, default=\"\")\n",
    "\n",
    "\n",
    "#data\n",
    "args.add_argument('--val_ratio', default=config['data']['val_ratio'], type=float)\n",
    "args.add_argument('--test_ratio', default=config['data']['test_ratio'], type=float)\n",
    "args.add_argument('--lag', default=config['data']['lag'], type=int)\n",
    "args.add_argument('--horizon', default=config['data']['horizon'], type=int)\n",
    "args.add_argument('--num_nodes', default=config['data']['num_nodes'], type=int)\n",
    "args.add_argument('--tod', default=config['data']['tod'], type=eval)\n",
    "args.add_argument('--normalizer', default=config['data']['normalizer'], type=str)\n",
    "args.add_argument('--column_wise', default=config['data']['column_wise'], type=eval)\n",
    "args.add_argument('--default_graph', default=config['data']['default_graph'], type=eval)\n",
    "#model\n",
    "args.add_argument('--model_type', default=config['model']['type'], type=str)\n",
    "args.add_argument('--g_type', default=config['model']['g_type'], type=str)\n",
    "args.add_argument('--input_dim', default=config['model']['input_dim'], type=int)\n",
    "args.add_argument('--output_dim', default=config['model']['output_dim'], type=int)\n",
    "args.add_argument('--embed_dim', default=config['model']['embed_dim'], type=int)\n",
    "args.add_argument('--hid_dim', default=config['model']['hid_dim'], type=int)\n",
    "args.add_argument('--hid_hid_dim', default=config['model']['hid_hid_dim'], type=int)\n",
    "args.add_argument('--num_layers', default=config['model']['num_layers'], type=int)\n",
    "args.add_argument('--cheb_k', default=config['model']['cheb_order'], type=int)\n",
    "args.add_argument('--solver', default='rk4', type=str)\n",
    "\n",
    "#train\n",
    "args.add_argument('--loss_func', default=config['train']['loss_func'], type=str)\n",
    "args.add_argument('--seed', default=config['train']['seed'], type=int)\n",
    "args.add_argument('--batch_size', default=config['train']['batch_size'], type=int)\n",
    "args.add_argument('--epochs', default=config['train']['epochs'], type=int)\n",
    "args.add_argument('--lr_init', default=config['train']['lr_init'], type=float)\n",
    "args.add_argument('--weight_decay', default=config['train']['weight_decay'], type=eval)\n",
    "args.add_argument('--lr_decay', default=config['train']['lr_decay'], type=eval)\n",
    "args.add_argument('--lr_decay_rate', default=config['train']['lr_decay_rate'], type=float)\n",
    "args.add_argument('--lr_decay_step', default=config['train']['lr_decay_step'], type=str)\n",
    "args.add_argument('--early_stop', default=config['train']['early_stop'], type=eval)\n",
    "args.add_argument('--early_stop_patience', default=config['train']['early_stop_patience'], type=int)\n",
    "args.add_argument('--grad_norm', default=config['train']['grad_norm'], type=eval)\n",
    "args.add_argument('--max_grad_norm', default=config['train']['max_grad_norm'], type=int)\n",
    "args.add_argument('--teacher_forcing', default=False, type=bool)\n",
    "#args.add_argument('--tf_decay_steps', default=2000, type=int, help='teacher forcing decay steps')\n",
    "args.add_argument('--real_value', default=config['train']['real_value'], type=eval, help = 'use real value for loss calculation')\n",
    "\n",
    "args.add_argument('--missing_test', default=False, type=bool)\n",
    "args.add_argument('--missing_rate', default=0.1, type=float)\n",
    "\n",
    "#test\n",
    "args.add_argument('--mae_thresh', default=config['test']['mae_thresh'], type=eval)\n",
    "args.add_argument('--mape_thresh', default=config['test']['mape_thresh'], type=float)\n",
    "args.add_argument('--model_path', default='', type=str)\n",
    "#log\n",
    "args.add_argument('--log_dir', default='../runs', type=str)\n",
    "args.add_argument('--log_step', default=config['log']['log_step'], type=int)\n",
    "args.add_argument('--plot', default=config['log']['plot'], type=eval)\n",
    "args.add_argument('--tensorboard',action='store_true',help='tensorboard')\n",
    "\n",
    "args = args.parse_args()\n",
    "init_seed(args.seed)\n",
    "\n",
    "GPU_NUM = args.device\n",
    "device = torch.device(f'cuda:{GPU_NUM}' if torch.cuda.is_available() else 'cpu')\n",
    "torch.cuda.set_device(device) # change allocation of current GPU\n",
    "\n",
    "print(args)\n",
    "\n",
    "#load dataset\n",
    "#train_loader, val_loader, test_loader, scaler, times = get_dataloader_cde(args,\n",
    "#                                                               normalizer=args.normalizer,\n",
    "#                                                               tod=args.tod, dow=False,\n",
    "#                                                               weather=False, single=False)\n",
    "\n",
    "#load raw st dataset\n",
    "data = load_st_dataset(args.dataset)        # B, N, D\n",
    "print(data.shape) \n",
    "# PEMSD4 timeframe: 01/2018 - 02/2018"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(datetime.date(2018, 1, 1), 'New year'),\n",
       " (datetime.date(2018, 1, 15), 'Birthday of Martin Luther King, Jr.'),\n",
       " (datetime.date(2018, 2, 19), \"Washington's Birthday\"),\n",
       " (datetime.date(2018, 5, 28), 'Memorial Day'),\n",
       " (datetime.date(2018, 7, 4), 'Independence Day'),\n",
       " (datetime.date(2018, 9, 3), 'Labor Day'),\n",
       " (datetime.date(2018, 10, 8), 'Columbus Day'),\n",
       " (datetime.date(2018, 11, 11), 'Veterans Day'),\n",
       " (datetime.date(2018, 11, 12), 'Veterans Day (Observed)'),\n",
       " (datetime.date(2018, 11, 22), 'Thanksgiving Day'),\n",
       " (datetime.date(2018, 11, 23), 'Thanksgiving Friday'),\n",
       " (datetime.date(2018, 12, 25), 'Christmas Day')]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from workalendar.usa.california import CaliforniaSanFrancisco\n",
    "holidays = [CaliforniaSanFrancisco().holidays(y) for y in [2018]][0]\n",
    "holidays"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sys.path.append(os.path.abspath(\"/home/rodr/code/mobility-baselines\"))\n",
    "from mobility_detrender import MobilityDetrender\n",
    "from evaluation import evaluation\n",
    "\n",
    "detrender = MobilityDetrender(day_duration=12*24, week_duration=7)\n",
    "trainset_len = len(data)-int(len(data)*(args.test_ratio+args.val_ratio)) # use only trainset for fitting HA\n",
    "detrender.fit(data[:trainset_len,:,0], start_date=datetime.date(2018,1,1), holidays=[0,14,49])\n",
    "detrender.visualize(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x108 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "loc = 0\n",
    "plt.figure(figsize=(18,1.5))\n",
    "img = plt.imshow(detrender.historical_avg[loc,:,:])\n",
    "plt.title(\"Historical average\")\n",
    "plt.xlabel(\"Time of day\")\n",
    "plt.ylabel(\"Day of week\")\n",
    "plt.colorbar(img)\n",
    "plt.savefig(\"stg-ncde-HA.pdf\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Residuals of Historical Average (on all data) -> MAPE: 0.159; MAE:  22.626; RMSE: 37.999\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(16992, 307)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "detrended_data = detrender.transform(data[:,:,0], start_date=datetime.date(2018,1,1), holidays=[0,14,49], mode='avg')\n",
    "evl_train = evaluation(data[:,:,0], detrender.ha_trend)\n",
    "print('Residuals of Historical Average (on all data) -> MAPE: %.3f; MAE:  %.3f; RMSE: %.3f' % tuple(evl_train)) \n",
    "detrended_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train:  (10173, 12, 307, 1) (10173, 12, 307, 1)\n",
      "Val:  (3375, 12, 307, 1) (3375, 12, 307, 1)\n",
      "Test:  (3375, 12, 307, 1) (3375, 12, 307, 1)\n",
      "Train HA:  (10173, 12, 307, 1) (10173, 12, 307, 1)\n",
      "Val HA:  (3375, 12, 307, 1) (3375, 12, 307, 1)\n",
      "Test HA:  (3375, 12, 307, 1) (3375, 12, 307, 1)\n"
     ]
    }
   ],
   "source": [
    "#split dataset by days or by ratio\n",
    "data_train, data_val, data_test = split_data_by_ratio(data, args.val_ratio, args.test_ratio)\n",
    "ha_train, ha_val, ha_test = split_data_by_ratio(detrender.ha_trend[:,:,np.newaxis], args.val_ratio, args.test_ratio)\n",
    "\n",
    "#add time window\n",
    "single=False\n",
    "x_tra, y_tra = Add_Window_Horizon(data_train, args.lag, args.horizon, single)\n",
    "x_val, y_val = Add_Window_Horizon(data_val, args.lag, args.horizon, single)\n",
    "x_test, y_test = Add_Window_Horizon(data_test, args.lag, args.horizon, single)\n",
    "print('Train: ', x_tra.shape, y_tra.shape)\n",
    "print('Val: ', x_val.shape, y_val.shape)\n",
    "print('Test: ', x_test.shape, y_test.shape)\n",
    "x_tra_ha, y_tra_ha = Add_Window_Horizon(ha_train, args.lag, args.horizon, single)\n",
    "x_val_ha, y_val_ha = Add_Window_Horizon(ha_val, args.lag, args.horizon, single)\n",
    "x_test_ha, y_test_ha = Add_Window_Horizon(ha_test, args.lag, args.horizon, single)\n",
    "print('Train HA: ', x_tra_ha.shape, y_tra_ha.shape)\n",
    "print('Val HA: ', x_val_ha.shape, y_val_ha.shape)\n",
    "print('Test HA: ', x_test_ha.shape, y_test_ha.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation set -> MAPE: 0.153; MAE:  24.431; RMSE: 40.938\n",
      "Test set -> MAPE: 0.17066; MAE:  26.259; RMSE: 42.861\n"
     ]
    }
   ],
   "source": [
    "evl_val = evaluation(y_val[:,0,:,:], y_val_ha[:,0,:,:])\n",
    "print('Validation set -> MAPE: %.3f; MAE:  %.3f; RMSE: %.3f' % tuple(evl_val)) \n",
    "\n",
    "evl_test = evaluation(y_test[:,0,:,:], y_test_ha[:,0,:,:])\n",
    "print('Test set -> MAPE: %.5f; MAE:  %.3f; RMSE: %.3f' % tuple(evl_test)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "50\n",
      "100\n",
      "150\n",
      "200\n",
      "250\n",
      "300\n",
      "(3375, 307, 12)\n",
      "(3375, 307, 12)\n",
      "Test set -> MAPE: 0.13387; MAE:  20.032; RMSE: 32.729\n"
     ]
    }
   ],
   "source": [
    "all_preds = []\n",
    "all_trues = []\n",
    "for loc in range(data.shape[1]):\n",
    "    if not loc % 50:\n",
    "        print(loc)\n",
    "    \n",
    "    loc_preds = []\n",
    "    loc_trues = []\n",
    "    for steps_ahead in range(12):\n",
    "        # prepare input data\n",
    "        x = x_tra[:, :, loc, 0] - x_tra_ha[:, :, loc, 0] # detrend input data for training\n",
    "        y = y_tra[:, steps_ahead, loc, 0] - y_tra_ha[:, steps_ahead, loc, 0] # detrend targets for training\n",
    "        \n",
    "        # fit ML models\n",
    "        reg = LinearRegression()\n",
    "        reg.fit(x, y)\n",
    "        \n",
    "        # predict on test set\n",
    "        trues = y_test[:, steps_ahead, loc, 0] # targets for test set (i.e., actual observations w/o detrending)\n",
    "        preds = reg.predict(x_test[:, :, loc, 0] - x_test_ha[:, :, loc, 0]) # make predictions on detrended inputs\n",
    "        preds = preds + y_test_ha[:, steps_ahead, loc, 0] # convert preds back to original scale (revert detrending)\n",
    "\n",
    "        loc_preds.append(preds)\n",
    "        loc_trues.append(trues)\n",
    "\n",
    "    all_preds.append(loc_preds)\n",
    "    all_trues.append(loc_trues)\n",
    "\n",
    "all_preds = np.array(all_preds).transpose([2,0,1])\n",
    "print(all_preds.shape)\n",
    "all_trues = np.array(all_trues).transpose([2,0,1])\n",
    "print(all_trues.shape)\n",
    "\n",
    "evl_test = evaluation(all_trues, all_preds)\n",
    "print('Test set -> MAPE: %.5f; MAE:  %.3f; RMSE: %.3f' % tuple(evl_test)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "50\n",
      "100\n",
      "150\n",
      "200\n",
      "250\n",
      "300\n",
      "(3375, 307, 12)\n",
      "(3375, 307, 12)\n",
      "Test set -> MAPE: 0.171; MAE:  26.264; RMSE: 42.868\n"
     ]
    }
   ],
   "source": [
    "# sanity check by having an empty model that predicts always zero - this is the same as predicting using HA only\n",
    "\n",
    "all_preds = []\n",
    "all_trues = []\n",
    "for loc in range(data.shape[1]):\n",
    "    if not loc % 50:\n",
    "        print(loc)\n",
    "    \n",
    "    loc_preds = []\n",
    "    loc_trues = []\n",
    "    for steps_ahead in range(12):\n",
    "        # prepare input data\n",
    "        x = x_tra[:, :, loc, 0] - x_tra_ha[:, :, loc, 0] # detrend input data for training\n",
    "        y = y_tra[:, steps_ahead, loc, 0] - y_tra_ha[:, steps_ahead, loc, 0] # detrend targets for training\n",
    "        \n",
    "        # no ML model\n",
    "        \n",
    "        # predict on test set\n",
    "        trues = y_test[:, steps_ahead, loc, 0] # targets for test set (i.e., actual observations w/o detrending)\n",
    "        preds = 0 # prediction is always zero\n",
    "        preds = preds + y_test_ha[:, steps_ahead, loc, 0] # convert preds back to original scale (revert detrending)\n",
    "\n",
    "        loc_preds.append(preds)\n",
    "        loc_trues.append(trues)\n",
    "\n",
    "    all_preds.append(loc_preds)\n",
    "    all_trues.append(loc_trues)\n",
    "\n",
    "all_preds = np.array(all_preds).transpose([2,0,1])\n",
    "print(all_preds.shape)\n",
    "all_trues = np.array(all_trues).transpose([2,0,1])\n",
    "print(all_trues.shape)\n",
    "\n",
    "evl_test = evaluation(all_trues, all_preds)\n",
    "print('Test set -> MAPE: %.3f; MAE:  %.3f; RMSE: %.3f' % tuple(evl_test)) "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}