Part-2.ipynb

Part-2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c5fcc709",
   "metadata": {},
   "source": [
    "### Creating dataset and dataloaders"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c960e129",
   "metadata": {},
   "outputs": [],
   "source": [
    "max_prediction_length = 6\n",
    "max_encoder_length = 24\n",
    "training_cutoff = data[\"time_idx\"].max() - max_prediction_length\n",
    "\n",
    "training = TimeSeriesDataSet(\n",
    "    data[lambda x: x.time_idx <= training_cutoff],\n",
    "    time_idx=\"time_idx\",\n",
    "    target=\"volume\",\n",
    "    group_ids=[\"agency\", \"sku\"],\n",
    "    min_encoder_length=max_encoder_length // 2,  # keep encoder length long (as it is in the validation set)\n",
    "    max_encoder_length=max_encoder_length,\n",
    "    min_prediction_length=1,\n",
    "    max_prediction_length=max_prediction_length,\n",
    "    static_categoricals=[\"agency\", \"sku\"],\n",
    "    static_reals=[\"avg_population_2017\", \"avg_yearly_household_income_2017\"],\n",
    "    time_varying_known_categoricals=[\"special_days\", \"month\"],\n",
    "    variable_groups={\"special_days\": special_days},  # group of categorical variables can be treated as one variable\n",
    "    time_varying_known_reals=[\"time_idx\", \"price_regular\", \"discount_in_percent\"],\n",
    "    time_varying_unknown_categoricals=[],\n",
    "    time_varying_unknown_reals=[\n",
    "        \"volume\",\n",
    "        \"log_volume\",\n",
    "        \"industry_volume\",\n",
    "        \"soda_volume\",\n",
    "        \"avg_max_temp\",\n",
    "        \"avg_volume_by_agency\",\n",
    "        \"avg_volume_by_sku\",\n",
    "    ],\n",
    "    target_normalizer=GroupNormalizer(\n",
    "        groups=[\"agency\", \"sku\"], transformation=\"softplus\"\n",
    "    ),  # use softplus and normalize by group\n",
    "    add_relative_time_idx=True,\n",
    "    add_target_scales=True,\n",
    "    add_encoder_length=True,\n",
    ")\n",
    "\n",
    "# create validation set (predict=True) which means to predict the last max_prediction_length points in time\n",
    "# for each series\n",
    "validation = TimeSeriesDataSet.from_dataset(training, data, predict=True, stop_randomization=True)\n",
    "\n",
    "# create dataloaders for model\n",
    "batch_size = 128  # set this between 32 to 128\n",
    "train_dataloader = training.to_dataloader(train=True, batch_size=batch_size, num_workers=0)\n",
    "val_dataloader = validation.to_dataloader(train=False, batch_size=batch_size * 10, num_workers=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "df575038",
   "metadata": {},
   "source": [
    "### Creating baseline model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dda95fef",
   "metadata": {},
   "source": [
    "Evaluating a Baseline model that predicts the next 6 months by simply repeating the last observed volume gives us a simle benchmark that we want to outperform."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f41d36fa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "293.0088195800781"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# calculate baseline mean absolute error, i.e. predict next value as the last available value from the history\n",
    "actuals = torch.cat([y for x, (y, weight) in iter(val_dataloader)])\n",
    "baseline_predictions = Baseline().predict(val_dataloader)\n",
    "(actuals - baseline_predictions).abs().mean().item()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9 (pytorch)",
   "language": "python",
   "name": "pytorch"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}