Custom metrics

This guide will show you how to automatically add custom metrics to your model from code using the Python SDK.

For more information on the syntax of custom metrics, check the Custom Metric Syntax documentation.

Defining Custom Metrics

To add new custom metrics:

conversion_rate = aporia.CustomMetric(
    "Conversion Rate",
    code="""
        count(filter="purchased=TRUE") / count()
    """,
)

model_revenue = aporia.CustomMetric(
    "Model Revenue",
    code="""
        sum(
            column="candidate_item_price",
            filter="purchased=TRUE AND ranking_index<=4"
        ) / count()*2
    """,
)

In this example, we're adding two new custom metrics - conversion rate and model revenue. To add the custom metrics to your model, pass them to the model object:

model = aporia.Model(
    "My Model",
    type=aporia.ModelType.RANKING,
    versions=[model_version],
    segments=[platform_segment, country_segments],
    custom_metrics=[conversion_rate, model_revenue]
)