Stat 565

Homework 4

Due in class Feb 4


Simulate an AR(1) process (you pick the parameter value) of length 30. To this simulated series, fit AR models of order 0 through 6. Which model has the smallest AIC?

Repeat the simulation 500 times, how often is the model with the lowest AIC the true model?

Repeat again with a longer time series, do things change?

Repeat with a different parameter value, do things change?


  1. Find and fit an ARIMA model to the deere2 dataset:

  2. Find and fit an ARIMA model to the robot dataset:


You should include:

  • how you chose the degrees of differencing d
  • how you chose p, q
  • diagnostics of your model fit
  • the estimates and standard errors of your model parameters