A decision tree regression model produces at most as many possible predictions as it has terminal nodes/leaves. A well fit decision tree will have been limited in the number of terminal nodes/leaves, to prevent overfitting. A histogram of predicted values will have at most the same number of bins with non-zero values, while certain bins within the range of the target variable will have no predictions, creating a “step” like pattern that will not appear smooth.

do you know what this line means?



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.