classifiers.py 文件源码

def train(self, datadict, labels=None):
        '''
        Runs the classifier training using the dictionary of label, features
        @param datadict: dictonary of label, features
        @param labels: (optional) list of labels. If given the order of labels is used from this list.
        '''

        # Set labels from data dict
        if labels is None:
            self.labels = datadict.keys()
        else:
            self.labels = labels
        # Train the GMM for BoF computation
        if self.model.gmm is None:
            print >> sys.stderr, 'Model not trained yet.'
            self.model.train(datadict, self.labels)

        print >> sys.stderr,'Computing',self.model.__class__.__name__,'...'
        # Parse dictionary into BoF representations and labels
        bofs, bofl = self._parse_dict(datadict, self.labels)

        #Create Multinomial Bayes
        print >> sys.stderr,'Training Multinomial Bayes ...'
        self.bay = bayes.MultinomialNB(alpha=0.5, fit_prior=False)
        self.bay.fit(bofs, bofl)
        return