SARIMAX - Forecasting weekly data

sarimax_fcast_weekly.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Forecasts\n",
      "2001-10-28/2001-11-03   -5.533356\n",
      "2001-11-04/2001-11-10   -2.526965\n",
      "2001-11-11/2001-11-17    0.475321\n",
      "Freq: W-SAT, dtype: float64\n",
      "\n",
      "Actual data\n",
      "2001-10-28/2001-11-03   -5.533356\n",
      "2001-11-04/2001-11-10   -2.526965\n",
      "2001-11-11/2001-11-17    0.475321\n",
      "Freq: W-SAT, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import statsmodels.api as sm\n",
    "\n",
    "from scipy.signal import lfilter\n",
    "\n",
    "# Simulate some data\n",
    "nobs = 100\n",
    "rs = np.random.RandomState(seed=1234)\n",
    "# (I made the shocks really small so that it'd be obvious\n",
    "# that SARIMAX was forecasting the correct periods)\n",
    "eps = lfilter([1], [1, -0.5], rs.normal(scale=1e-10, size=nobs))\n",
    "exog = rs.normal(size=(nobs, 2))\n",
    "beta = np.array([5, -2])\n",
    "endog = exog.dot(beta) + eps\n",
    "\n",
    "# Set the data be be weekly, anchored to Saturday\n",
    "index = pd.PeriodIndex(start='2000-01', periods=nobs, freq='W-SAT')\n",
    "endog = pd.Series(endog, index=index)\n",
    "exog = pd.DataFrame(exog, index=index)\n",
    "\n",
    "# Setup the dates we want to estimate the model on and forecast until\n",
    "end_estimation = pd.Period('2001-10-27', freq=endog.index.freq)\n",
    "end_forecast = pd.Period('2001-11-17', freq=endog.index.freq)\n",
    "\n",
    "# Setup the model and just smooth at the correct parameters\n",
    "# so that we can check how forecast is doing\n",
    "mod = sm.tsa.SARIMAX(endog.loc[:end_estimation], exog.loc[:end_estimation], order=(1, 0, 0))\n",
    "res = mod.smooth([5, -2, 0.5, 1e-10])\n",
    "\n",
    "# Make the forecasts\n",
    "print('Forecasts')\n",
    "print(res.forecast(end_forecast, exog=exog.loc[end_estimation + 1:end_forecast]))\n",
    "print('\\nActual data')\n",
    "print(endog.loc[end_estimation + 1:end_forecast])"
   ]
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}