MakeHierarchicalResponse

OK = MakeHierarchicalResponse(L, MLE) builds a two-stage model from the response feature models in L and optionally runs maximum likelihood estimation, MLE.

If there are more response features than the number of parameters in the local model, the software chooses the subset of response features that leads to the best hierarchical response. The best hierarchical response is chosen using PRESS RMSE if all the response feature models are linear. Otherwise, the best hierarchical response is chosen using two-stage RMSE. For more information, see PRESS statistic.

MakeHierarchicalResponse performs a similar function to ChooseAsBest for response models. You can call MakeHierarchicalResponse directly or indirectly by calling CreateAlternativeModels for a local model. If you call CreateAlternativeModels for a local model, MakeHierarchicalResponse is called automatically.

If the local and response models are not ready to calculate a two-stage model, the software issues an error. This situation can occur if you have created alternative models and not chosen the best. You must select a sufficient number of response features models to calculate the two-stage model.