MATLAB: How we can test the Ensemble??..

pattern recognition

I can't get one thing…. As in 'fitensemble' function, we just create the ensemble 'ens' on training data and generate the class models of learners… Now, how we will use that ensemble on validation set??.. and how will check its performance?? i.e it is performing well or poor on validation set and then on test set finally??

Best Answer

http://www.mathworks.com/help/toolbox/stats/bsvjye9.html#bsvjyrp

Related Solutions

MATLAB: How to get prediction scores from exported Classification Learner SVM model

Hello Ashley,

The predict function expects an input classifier of type such as ClassificationKNN or ClassificationTree etc. In this case you are passing a variable T which is if type struct. That is the reason for the error.

Type the following at the command prompt to see the contents of the struct:

>> T %Variable name of the exported classifier struct

Here one of the fields will be the classification object. For e.g. for my exported model referred by struct variable trainedClassifier, I see the following.

>> trainedClassifier
trainedClassifier = 
           predictFcn: [function_handle]
    RequiredVariables: {1x11 cell}
    ClassificationKNN: [1x1 ClassificationKNN]
                About: 'This struct is a tra…'
         HowToPredict: 'To make predictions …'

Note here that one of the fields is the actual classification object, ClassificationKNN. Note that, based on the type of trained classifier this might differ. To access this we can use the dot operator.

kNNMdl = trainedClassifier.ClassificationKNN;

Now use this as input to the predict function to get scores:

[labels,score] = predict(kNNMdl,ValData);

MATLAB: How to cross validate the data and use it for ensemble

Computing loss on test data and computing loss by cross-validation are two separate tasks. To compute loss on test data, you need to train an ensemble using all training data you have. To compute loss by say 10-fold validation, you need to grow 10 ensembles, each on 9/10 of your training data and then average loss over the left-out 1/10 parts. I am not sure why you expect that both tasks would be handled by one object. Even if they were handled by one object, it would not buy you anything in terms of CPU time or memory. Just do two separate things: 1) Grow an ensemble on all training data and use it to compute the test loss, and 2) Cross-validate this ensemble using its crossval method and use the kfoldLoss method of the partitioned ensemble (new object) to compute the cross-validated loss.

Best Answer

Related Solutions

MATLAB: How to get prediction scores from exported Classification Learner SVM model

MATLAB: How to cross validate the data and use it for ensemble

Related Question