datadriftconfig.yaml

---
- database_config:
    database_type: postgres
    username: 'super_admin'
    pwd: 'SomeSecretPassword'
    host: 'postgres_test'
    port: 5432
    database: 'postgres'
    tables:
      - table_name: 'soup_shop'
        date_column: 'date'
        target_column: 'price'
        numerical_columns:
            - 'temperature'
        categorical_columns:
            - 'flavor'
      - table_name: 'weather'
        date_column: 'date'
        target_column: 'precipitation'
        numerical_columns:
            - 'temperature'
        categorical_columns:
            - 'season'
- drift_config:
#    sample_size: 1000 Use to pull subsample of data if data is too large
    reports: # DataQualityPreset, DataDriftPreset, TargetDriftPreset, RegressionPerformancePreset, ClassificationPerformancePreset
        - report_type: 'DataDriftPreset' # parameters: columns, stattest, cat_stattest, num_stattest, per_column_stattest, stattest_threshold, cat_stattest_threshold, num_stattest_threshold, per_column_stattest_threshold, drift_share
          parameters:
            drift_share: 0.75
#        - report_type: 'DataQualityPreset' # parameters: columns - Can not be used for multiple table configurations
#          parameters:
#            columns: []
#        - report_type: 'TargetDriftPreset' # parameters: columns, stattest, cat_stattest, num_stattest, per_column_stattest, stattest_threshold, cat_stattest_threshold, num_stattest_threshold, per_column_stattest_threshold
#          parameters:
#            columns: []
#        - report_type: 'RegressionPerformancePreset' # parameters: columns
#          parameters:
#            columns: []
#        - report_type: 'ClassificationPerformancePreset' # parameters: columns, probas_threshold, k
#          parameters:
#            columns: []
#            probas_threshold: 0.5
#            k: 5