liveBook · Manning

Metric preset

As explained in Machine Learning Platform Engineering

A metric preset is a prebuilt Evidently report definition that aggregates metrics for a specific analysis use case.

A metric preset is a prebuilt Evidently report definition that aggregates measurements for a specific analysis use case. It acts as a reusable template for production monitoring, such as data drift, data quality, or regression performance. section 6.2.1

Common presets include:

  • DataDriftPreset compares feature distributions between reference and current datasets. section 6.2.1
  • DataQualityPreset aggregates data-quality measurements. section 6.2.1
  • RegressionPreset aggregates regression-performance measurements. section 6.2.1

Generate a data-drift report

Define the preset, run it with reference and current data, and save the rendered report:\n\nreport = Report(metrics=[DataDriftPreset()])\nreport.run(reference_data=reference_data, current_data=current_data)\nreport.save_html("drift_report.html")

How Machine Learning Platform Engineering covers it

How deep it goes
Defines presets, shows report execution and HTML export, and illustrates drift scores without detailing preset internals.
Examples
Uses DataDriftPreset to compare feature distributions and save an HTML report.
What it assumes
Assumes reference and current datasets plus familiarity with metrics and reports.
What you can do after
Apply it via Evidently to generate reusable monitoring reports from reference and current datasets.
How it connects
Connects metrics with reference datasets, current datasets, data drift, data quality, and regression performance.
Cover of Machine Learning Platform Engineering
liveBookmanning.com
Read Machine Learning Platform Engineering on liveBook Get the book at manning.com

This explainer was generated by Manning's knowledge graph from the text of Machine Learning Platform Engineering. The section links point to the parts of the book it draws on.