Classification crossval results key error

[bluetyson]

I get an error

File "c:\users\rscott\appdata\local\continuum\anaconda3\envs\gdaltest\lib\site-packages\uncoverml\validate.py", line 237, in export_crossval
    to_text = [self.y_true, self.y_pred['Prediction'], self.positions[:,0], self.positions[:,1]]
KeyError: 'Prediction'

Looking at that dictionary (for a 24 class problem)

dict_keys(['most_likely', '0.0_0', '1.0_1', '2.0_2', '3.0_3', '4.0_4', '5.0_5', '6.0_6', '7.0_7', '8.0_8', '9.0_9', '10.0_10', '11.0_11', '12.0_12', '13.0_13', '14.0_14', '15.0_15', '16.0_16', '17.0_17', '18.0_18', '19.0_19', '20.0_20', '21.0_21', '22.0_22', '23.0_23'])

is the most_likely supposed to go there in the case of classification, if going to look the same as regression? e.g. using self.classification?

[RichardScottOZ]

        # Make sure we convert numpy arrays to lists
        scores = {s: v if np.isscalar(v) else v.tolist()
                  for s, v in self.scores.items()}

        with open(config.crossval_scores_file, 'w') as f:
            json.dump(scores, f, sort_keys=True, indent=4)

        if self.classification:
            to_text = [self.y_true, self.y_pred['most_likely'], self.positions[:,0], self.positions[:,1]]
        else:
            to_text = [self.y_true, self.y_pred['Prediction'], self.positions[:,0], self.positions[:,1]]
        #print(self.y_pred.keys())
        #print("classification flag", self.classification)
        #to_text = [self.y_true, self.y_pred, self.positions[:,0], self.positions[:,1]]

        np.savetxt(config.crossval_results_file, X=np.vstack(to_text).T, 
                   delimiter=',', fmt='%.4f', header='y_true,y_pred,x,y')

        if os.path.exists(config.raw_covariates) and os.path.exists(config.raw_covariates_mask):
            # Also add prediction values to rawcovariates.csv - yes this file 
            # is very overloaded and we need to fix the output situation.
            # Get indicies sorted by location so we can insert the 
            # prediction in the correct row.
            inds = np.lexsort(self.positions.T)
            #sorted_predictions = self.y_pred['Prediction'][inds]
            if self.classification:
                sorted_predictions = self.y_pred['most_likely'][inds]
            else:
                sorted_predictions = self.y_pred['Prediction'][inds]

[RichardScottOZ]

This is still a problem it would seem?