Forecasting with an AR Model
In addition to estimating the parameters of a model that you did in the last exercise, you can also do forecasting, both in-sample and out-of-sample using statsmodels. The in-sample is a forecast of the next data point using the data up to that point, and the out-of-sample forecasts any number of data points in the future. You can plot the forecasted data using the function plot_predict(). You supply the starting point for forecasting and the ending point, which can be any number of data points after the data set ends.
For the simulated data in DataFrame simulated_data_1, with \(\small \phi=0.9\), you will plot out-of-sample forecasts and confidence intervals around those forecasts.
This is a part of the course
“Time Series Analysis in Python”
Exercise instructions
- Import the class
ARIMAand also import the functionplot_predict - Create an instance of the
ARIMAclass calledmodusing the simulated data in DataFramesimulated_data_1and the order (p,d,q) of the model (in this case, for an AR(1)),order=(1,0,0) - Fit the model
modusing the method.fit()and save it in a results object calledres - Plot the in-sample data starting with data point 950
- Plot out-of-sample forecasts of the data and confidence intervals using the
plot_predict()function, starting where the data ends at the 1000th point, and ending the forecast at point 1010
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Import the ARIMA and plot_predict from statsmodels
from statsmodels.tsa.arima.model import ARIMA
from statsmodels.graphics.tsaplots import plot_predict
# Forecast the first AR(1) model
mod = ARIMA(___, order=___)
res = mod.fit()
# Plot the data and the forecast
fig, ax = plt.subplots()
simulated_data_1.loc[950:].plot(ax=ax)
plot_predict(res, start=___, end=___, ax=ax)
plt.show()This exercise is part of the course
Time Series Analysis in Python
In this four-hour course, you’ll learn the basics of analyzing time series data in Python.
In this chapter you'll learn about autoregressive, or AR, models for time series. These models use past values of the series to predict the current value.
Exercise 1: Describe AR ModelExercise 2: Simulate AR(1) Time SeriesExercise 3: Compare the ACF for Several AR Time SeriesExercise 4: Match AR Model with ACFExercise 5: Estimating and Forecasting AR ModelExercise 6: Estimating an AR ModelExercise 7: Forecasting with an AR ModelExercise 8: Let's Forecast Interest RatesExercise 9: Compare AR Model with Random WalkExercise 10: Choosing the Right ModelExercise 11: Estimate Order of Model: PACFExercise 12: Estimate Order of Model: Information CriteriaWhat is DataCamp?
Learn the data skills you need online at your own pace—from non-coding essentials to data science and machine learning.