H2O Model Validation's drift detection test is a powerful tool for detecting changes in the distribution of variables in a model's input data, which can lead to performance degradation if not addressed. The test uses two datasets captured at different times to assess how data has changed over time and calculates two metrics, Population Stability Index (PSI) and drift Score, to measure the degree of change.
Variables with a higher PSI or drift score indicate a higher drift, which can lead to a decrease in model performance. Important variables in a model with a high score may require model retraining to maintain accuracy.
H2O Model Validation offers several settings for configuring the drift detection test.
By using the drift detection test in H2O Model Validation, data scientists and machine learning engineers can proactively identify changes in their data and take appropriate actions to maintain the accuracy of their models.
- To learn how to create a drift detection test, see Create a drift detection test.
- See Settings: Drift detection to learn about all the settings for a drift detection validation test.
- See Metrics: Drift detection to learn about all the metrics for a drift detection validation test.