Bayesian optimization with fixed 'Method' for fitcensemble

I want to optimize an ensemble classification model with Bayesian optimization.

cvpt = cvpartition(Y_train , "KFold" , 5)

opt = struct("CVPartition" , cvpt , "MaxObjectiveEvaluations" , 20)

Mdl = fitcensemble(X_train , Y_train , "OptimizeHyperparameters", "auto" , "HyperparameterOptimizationOptions" , opt)

However, I want to use 'AdaBoostM2' as 'Method' and optimize all the other hyperparameters. How do I specify this?

Answer 1

[Alan Weiss]

Mdl = fitcensemble(X_train , Y_train ,...
    "OptimizeHyperparameters", {'NumLearningCycles','LearnRate'} ,...
    "HyperparameterOptimizationOptions" , opt,...
    "Method","AdaBoostM2")

Alan Weiss

MATLAB mathematical toolbox documentation