function to score a bunch of instances together for multilabel revscoring

score_many.py

    def score_many(self, feature_values):                                                                                                                                
        # Re-vectorize features -- this expands/flattens sub-FeatureVectors                                                                                              
        fv_vectors = [vectorize_values(fv) for fv in feature_values]                                                                                                     
                                                                                                                                                                         
        # Scale and transform (if applicable)                                                                                                                            
        scaled_fv_vectors = self.fit_scaler_and_transform(fv_vectors)                                                                                                    
        predictions = self.estimator.predict(scaled_fv_vectors)                                                                                                          
        #probas = self.estimator.predict_proba(scaled_fv_vectors)                                                                                                        
        probabilities = []                                                                                                                                               
        docs = []                                                                                                                                                        
        if self.multilabel:                                                                                                                                              
            #probas = self.estimator.predict_proba(scaled_fv_vectors)                                                                                                    
            prob_matrix = np.empty((0, len(feature_values)))                                                                                                             
            probas = self.estimator.predict_proba(scaled_fv_vectors)                                                                                                     
            for label_prob in probas:                                                                                                                                    
                curr_label_prob = []                                                                                                                                     
                for prob in label_prob:                                                                                                                                  
                    curr_label_prob.append(prob[1])                                                                                                                      
                prob_matrix = np.append(prob_matrix, [curr_label_prob], axis=0)                                                                                          
            # This converts probability matrix to [n_samples, n_labels] for ease                                                                                         
            # of iteration                                                                                                                                               
            prob_matrix = np.transpose(prob_matrix)                                                                                                                      
            for pr in prob_matrix:                                                                                                                                       
                probabilities.append({label: proba                                                                                                                       
                                     for label, proba in zip(self.labels, pr)})                                                                                          
        else:                                                                                                                                                            
            labels = self.estimator.classes_                                                                                                                             
            probas = self.estimator.predict_proba(scaled_fv_vectors)                                                                                                     
            for prob in probas:                                                                                                                                          
                probabilities.append({label: prob                                                                                                                        
                                      for label, prob in zip(labels, prob)})                                                                                             
        for pred, prob in zip(predictions, probabilities):                                                                                                               
            preds = self.label_normalizer.denormalize(pred)                                                                                                              
            doc = {'prediction': preds, 'probability': prob}                                                                                                             
            docs.append(util.normalize_json(doc))                                                                                                                        
        return docs