Consider again the daily PM2.5 series of Beijing. Reserve the last 365 observations for out-of-sample prediction. Apply random forests to obtain predictions.

Use the same lagged values as in Example 4.6 as the predictors. Try several random forests and comment on their performance.