def _get_time_ranges()

in src/lookoutequipment/evaluation.py [0:0]

• 14 lines of code
• 8 McCabe index (conditional complexity)

    def _get_time_ranges(self):
        """
        Extract DateTimeIndex with normal values and anomalies from the
        predictions generated by the model.
        
        Returns:
            pandas.DateTimeIndex:
                Timestamp index for all the values marked as normal during the
                training period
            pandas.DateTimeIndex:
                Timestamp index for all the values predicted as anomalies by
                the model during the evaluation period
        """
        # Extract the first time series 
        tag = list(self.df_list.keys())[0]
        tag_df = self.df_list[tag]
        
        # Initialize the predictions dataframe:
        predictions_df = pd.DataFrame(columns=['Prediction'], index=tag_df.index)
        predictions_df['Prediction'] = 0

        # Loops through the predicted and labelled anomalies
        # ranges and set these predictions to 1 (predicted) 
        # or 2 (initially known):
        for index, row in self.predicted_ranges.iterrows():
            predictions_df.loc[row['start']:row['end'], 'Prediction'] = 1
        for index, row in self.labelled_ranges.iterrows():
            predictions_df.loc[row['start']:row['end'], 'Prediction'] = 2

        # Limits the analysis range to the evaluation period:
        predictions_df = predictions_df[self.training_start:self.evaluation_end]
        
        # Build a DateTimeIndex for normal values and anomalies:
        index_normal = predictions_df[predictions_df['Prediction'] == 0].index
        index_anomaly = predictions_df[predictions_df['Prediction'] == 1].index
        index_known = predictions_df[predictions_df['Prediction'] == 2].index
        
        return index_normal, index_anomaly, index_known