Results

After training, DSS computes regression metrics for each target and also displays global metrics.

Metrics

Multi-target regression model reports include both global metrics and per-target metrics.

Global metrics are computed by averaging the per-target metric values. For hyperparameter search and model comparisons in the Result screen, the optimized metric is the global version of the metric selected in Design > Metrics.

For custom metrics, DSS applies the custom metric function independently to each target, then uses the arithmetic mean across targets by default. A custom target aggregation function can override this default aggregation.

Per-target regression details

Multi-target regression exposes regression result details separately for each target. These details include residuals, scatter plot data, prediction distributions, and error statistics.

Use per-target details to identify targets for which the model performs differently, even when the global metrics look acceptable.

Feature importance

Model-provided feature importance is available when the algorithm exposes it, for example for Random Forest models.

Shapley feature importance is available for multi-target regression models. DSS computes Shapley explanations for each target. The global Shapley ranking is computed by normalizing the absolute importances for each target before averaging them across targets, so targets with larger numerical scales do not dominate the ranking.