1. Learn
    2. /
  2. Courses
    3. /
  3. Time Series Analysis in Python

Exercise

Estimate Order of Model: Information Criteria

Another tool to identify the order of a model is to look at the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). These measures compute the goodness of fit with the estimated parameters, but apply a penalty function on the number of parameters in the model. You will take the AR(2) simulated data from the last exercise, saved as simulated_data_2, and compute the BIC as you vary the order, p, in an AR(p) from 0 to 6.

Instructions

100 XP
  • Import the ARIMA module for estimating the parameters and computing BIC.
  • Initialize a numpy array BIC, which we will use to store the BIC for each AR(p) model.
  • Loop through order p for p = 0,…,6.
    • For each p, fit the data to an AR model of order p.
    • For each p, save the value of BIC using the .bic attribute (no parentheses) of res.
  • Plot BIC as a function of p (for the plot, skip p=0 and plot for p=1,…6).