Assessing how well a model performs is an essential part of applied analytics. In SPSS Statistics, model fit and classification accuracy procedures are often spread across different procedures, making it harder to navigate when you are trying to assess model performance using multiple criteria.
Model Fit Hub brings a practical set of model fit statistics and charts together in one place. It is designed as a lightweight “hub” dialog that works for both continuous outcomes (regression) and categorical outcomes (binary classification), and returns results as SPSS pivot tables and R charts.
What is Model Fit Hub?
Model Fit Hub is a custom dialog and extension command so that from a single dialog, you can request:
- Regression fit metrics (continuous outcomes): N, R-squared, MSE, RMSE, MAE, MAPE, and Max APE.
- Classification metrics (binary outcomes): Accuracy, Balanced accuracy, Precision (PPV), Recall/Sensitivity (TPR), Specificity (TNR), MCC, and F1.
- Probability-based diagnostics: ROC AUC, Gini, Log Loss, Brier score, and the KS statistic.
- Confusion matrix summary and heatmap: True/False positives and negatives shown as percentages with counts.
- Charts: ROC curve, Gains chart, and Lift chart (when a probability variable is provided).
- Optimal cutpoints (cross-validated): a table showing cut-offs that maximise selected classification metrics, with cross-validated performance estimates.
The dialog also generates SPSS-like syntax based on the custom STATS FITHUB command, allowing you to paste and reuse your settings in syntax files and automation scripts.
Why it’s useful
- Productivity: common fit measures and charts are available in a single location.
- Consistency: the same set of definitions and output formatting can be applied across projects and teams.
- Better threshold decisions: probability-based metrics and cross-validated optimal cutpoints help you choose cut-offs that match your objective.
- Free and self-contained: available as a single extension bundle with its own HTML help.
Installation
To use the tool, simply download the Model Fit Hub v4a.spe file and install it in SPSS Statistics by clicking Extensions > Install Local Extension Bundle. Once installed, the Model Fit Hub dialog will be available near the lower end of the Analyze > Regression menu.