gistfile1.txt

 115         # Check parameters                                                      
 116         self._validate_estimator()                                              
 117                                                                                 
 118         # Clear any previous fit results                                        
 119         self.estimators_ = []                                                   
 120         self.estimator_weights_ = np.zeros(self.n_estimators, dtype=np.float)   
 121         self.estimator_errors_ = np.ones(self.n_estimators, dtype=np.float)     
 122                                                                                 
 123         for iboost ∈ xrange(self.n_estimators):                                 
 124             # Boosting step                                                     
 125             sample_weight, estimator_weight, estimator_error = self._boost(     
 126                 iboost,                                                         
 127                 X, y,                                                           
 128                 sample_weight)                                                  
 129                                                                                 
 130             # Early termination                                                 
 131             if sample_weight is None:                                           
 132                 break                                                           
 133                                                                                 
 134             self.estimator_weights_[iboost] = estimator_weight                  
 135             self.estimator_errors_[iboost] = estimator_error                    
 136                                                                                 
 137             # Stop if error is zero                                             
 138             if estimator_error ≡ 0:                                             
 139                 break                                                           
 140                                                                                 
 141             sample_weight_sum = np.∑(sample_weight)                             
 142                                                                                 
 143             # Stop if the sum of sample weights has become non-positive         
 144             if sample_weight_sum ≤ 0:                                           
 145                 break                                                           
 146                                                                                 
 147             if iboost < self.n_estimators - 1:                                  
 148                 # Normalize                                                     
 149                 sample_weight /= sample_weight_sum                              
 150                                                                                 
 151         return self