Naive Bayes Posterior Probability

Hi,

I am using a Naive Bayes Classifier which I train with:

Mdl = fitcnb(data,target,...
    'Holdout',0.10,...
    'OptimizeHyperparameters','auto',...
    'HyperparameterOptimizationOptions',...
    struct('AcquisitionFunctionName','expected-improvement-plus'));

So when I predit the class of the data results in values of 0 and 1, because I only have two classes. This is pretty nice, but I need to know the certainty of this estimate. So I want to have something like:

class 1: probability 0.4

class 2: probability 0.6

or something like this.

Is the correct function for this the posterior function? As far as I understood it this function is only available for the deprecated NaiveBayes class.

Does anybody know how I can get the probability of the estimate?

Answer 1

[Don Mathis]

You can get the posterior probabilities from the second output of the predict method. See this page: http://www.mathworks.com/help/stats/compactclassificationnaivebayes.predict.html

[label,Posterior] = predict(Mdl,X)