What parameters are optimized by default when the crossval-on name-value pair option is used in the fitrensemble function?
1 view (last 30 days)
For eg, when the following command is used, what parameters/hyperparamters are validated by default when the crossval-on name-value pair option is used in the fitrensemble function?
t = templateTree('MaxNumSplits',1);
Mdl = fitrensemble(X,MPG,'Learners',t,'CrossVal','on');
Aditya Patil on 12 Jul 2021
Cross validation splits the data into K partitions. Then it trains the models on the K permutations of (K - 1) sets and validates it on the remaining 1 set. For example, if you use 10-fold validation, it will train on 9 different permutations of the sets, each having 9 sets for training, and 1 for validation.
As such, there is no dependence on the parameters of the model.