mlm_insights.builder.usage package¶

Submodules¶

class mlm_insights.builder.usage.insights_usage_manager.InsightsUsageManager¶

Bases: ABC

Computes the metrics based on the input schema

abstract select_metrics(input_schema: Dict[str, FeatureType]) → MetricDetail¶

class mlm_insights.builder.usage.minimal_insights_usage_manager.MinimalInsightsUsageManager¶

Selects metrics based on feature metadata based on the following criteria:

Following metrics are tracked according to the feature’s data type:

Int or float

Count

Min, Max, Sum

Median

Standard Deviation

Quartiles

Distribution

Cardinality

String

Count

Cardinality

Uniqueness

Frequent Items

We add all dataset level metrics by default