DataDriftMetricThreshold 類別
注意
這是實驗性類別,可以隨時變更。 請參閱 https://aka.ms/azuremlexperimental 以取得詳細資訊。
資料漂移計量閾值
- 繼承
-
azure.ai.ml.entities._monitoring.thresholds.MetricThresholdDataDriftMetricThreshold
建構函式
DataDriftMetricThreshold(*, data_type: ~typing.Literal[<MonitorFeatureType.CATEGORICAL: 'categorical'>, <MonitorFeatureType.NUMERICAL: 'numerical'>] = None, threshold: float = None, metric: str | None = None, numerical: ~azure.ai.ml.entities._monitoring.thresholds.NumericalDriftMetrics = None, categorical: ~azure.ai.ml.entities._monitoring.thresholds.CategoricalDriftMetrics = None)
參數
- applicable_feature_type
- Literal[ azure.ai.ml.constants.MonitorFeatureType.CATEGORICAL , azure.ai.ml.constants.MonitorFeatureType.MonitorFeatureType.NUMERICAL]
必要
計量閾值的功能類型
- metric_name
- Literal[ MonitorMetricName.JENSEN_SHANNON_DISTANCE , azure.ai.ml.constants.MonitorMetricName.NORMALIZED_WASSERSTEIN_DISTANCE , azure.ai.ml.constants.MonitorMetricName.POPULATION_STABILITY_INDEX , azure.ai.ml.constants.MonitorMetricName.TWO_SAMPLE_KOLMOGOROV_SMIRNOV_TEST , azure.ai.ml.constants.MonitorMetricName.PEARSONS_CHI_SQUARED_TEST]
必要
要計算的計量
意見反應
https://aka.ms/ContentUserFeedback。
即將登場:在 2024 年,我們將逐步淘汰 GitHub 問題作為內容的意見反應機制,並將它取代為新的意見反應系統。 如需詳細資訊,請參閱:提交並檢視相關的意見反應