I am practicing random forest regressions with monthly US inflation rates as sample data. I am predicting the inflation rate simply using its past month value. My last data point is for May 22 when monthly inflation was 0.97%, the mean for the whole sample is 0.29%. The predicted inflation rate for June is monotonically decreasing in the maximum tree depth. For example when I set it to 5, the prediction is 0.72% but it is merely 0.33% when I set the max depth to 20, which is only a bit above the unconditional mean. The linear OLS prediction would be 0.68%. Could someone explain the intuition behind this decline in the forecast value towards the mean?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.