Principle:Evidentlyai Evidently Drift Dataset Metrics
| Knowledge Sources | |
|---|---|
| Domains | ML_Monitoring, Data_Quality, Statistical_Testing |
| Last Updated | 2026-02-14 12:00 GMT |
Overview
Dataset-level drift and quality metrics that summarize drift detection and missing data across multiple columns simultaneously.
Description
Drift Dataset Metrics provide aggregate views of data quality across an entire dataset:
- DriftedColumnsCount: Counts how many columns have detected drift (above threshold) and the overall drift share
- MissingValueCount: Counts null/NaN values for a specific column
These complement the column-level ValueDrift metric by providing summary statistics that indicate overall dataset health.
Usage
Use alongside ValueDrift in drift monitoring Reports to get both per-column and aggregate drift statistics.
Theoretical Basis
Dataset-level drift aggregation:
# Pseudocode
drifted_count = sum(1 for col in columns if drift_test(col) > threshold)
drift_share = drifted_count / total_columns
# If drift_share > alarm_threshold: alert
Related Pages
Implemented By
Page Connections
Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment