evaluating-regression-models

What problem does it solve?

This Skill turns raw regression predictions into a defensible evaluation, so you can tell whether a model is actually useful instead of relying on a single headline score.

Core Features & Use Cases

• Complete scoring: Reports RMSE, MAE, R-squared, and adjusted R-squared when multiple features are involved.
• Residual diagnostics: Checks residuals versus fitted values, normality with a QQ assessment, and residual shape with a histogram.
• Validation discipline: Requires cross-validation with mean and spread across folds, and switches to time-series-aware validation when forecasting data is involved.
• Practical guardrails: Refuses to accept R-squared by itself, flags possible leakage, and highlights influential or high-error rows for review.

Quick Start

Ask Claude to evaluate the regression model by providing y_true, y_pred, the number of features, and any cross-validation or time-series context.