Data Drift
ML Concept
Overview
Use casedetecting changes in input data distributions that can degrade machine learning model performance
Knowledge graph stats
Claims12
Avg confidence89%
Avg freshness100%
Last updatedUpdated 4 days ago
Trust distribution
100% unverified
Governance
Not assessed
Data Drift
concept
The change in model input data over time that can negatively impact model performance.
Compare with...primary use case
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| detecting changes in input data distributions that can degrade machine learning model performance | ○Unverified | High | Fresh | 1 |
causes issue
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| model performance degradation | ○Unverified | High | Fresh | 1 |
category
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| ML monitoring and observability | ○Unverified | High | Fresh | 1 |
supported by platform
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| Amazon SageMaker Model Monitor | ○Unverified | High | Fresh | 1 |
| Google Cloud Vertex AI Model Monitoring | ○Unverified | Moderate | Fresh | 1 |
| Azure Machine Learning | ○Unverified | Moderate | Fresh | 1 |
mitigation strategy
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| model retraining | ○Unverified | Moderate | Fresh | 1 |
related to
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| concept drift | ○Unverified | Moderate | Fresh | 1 |
| covariate shift | ○Unverified | Moderate | Fresh | 1 |
detection method
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| statistical tests like Kolmogorov-Smirnov test | ○Unverified | Moderate | Fresh | 1 |
open source tool
| Value | Trust | Confidence | Freshness | Sources |
|---|---|---|---|---|
| Evidently AI | ○Unverified | Moderate | Fresh | 1 |
| Alibi Detect | ○Unverified | Moderate | Fresh | 1 |