Evaluate with model insights¶
Model insights help to interpret, explain, and validate what drives a model’s predictions. Using these tools can help to assess what to do in your next experiment. Available insights are dependent on experiment type, but may include the insights listed in the predictive modeling insights table below. Availability of sliced insights is also model-dependent. To see a model's insights, click on the model in the left-pane Leaderboard. Note that different insights are available for predictive experiments.
Available insights¶
Model insights help to interpret, explain, and validate what drives a model’s predictions. Available insights are dependent on experiment type, but may include the insights listed in the table below.
| Insight | Description | Problem type | Sliced insights? |
|---|---|---|---|
| Accuracy Over Time | Visualizes how predictions change over time. | Time-aware | |
| Anomaly Assessment | Plots data for the selected backtest and provides, below the visualization, SHAP explanations for up to 500 anomalous points. | Time series | |
| Anomaly Over Time | Visualizes where anomalies occur across the timeline of your data. | Time-aware | |
| Blueprint | Provides a graphical representation of data preprocessing and parameter settings. | All | |
| Coefficients | Provides a visual indicator of the relative effects of the 30 most important variables. | All; linear models only | |
| Compliance documentation | Generates individualized documentation to provide comprehensive guidance on what constitutes effective model risk management. | ||
| Feature Effects | Conveys how changes to the value of each feature change model predictions | All | ✔ |
| Feature Impact | Shows which features are driving model decisions. | All | ✔ |
| Forecasting Accuracy | Depicts how well a model predicts at each forecast distance in the experiment's forecast window. | Time series | |
| Forecast vs Actual | Predicts multiple values for each point in time (forecast distances). | Time series | |
| Lift Chart | Depicts how well a model segments the target population and how capable it is of predicting the target. | All | ✔ |
| Period Accuracy | Shows model performance over periods within the training dataset. | Time-aware | |
| ROC Curve | Provides tools for exploring classification, performance, and statistics related to a model. | Classification | ✔ |
| Series Insights | Provides series-specific information for multiseries experiments. | Time series | |
| Stability | Provides a summary of how well a model performs on different backtests. | Time-aware |
What's next?¶
After selecting a model, you can, from within the experiment: