gaQSAR 1.2.3
Improvements
- Improved Williams plot diagnostics and documentation for double
cross-validation.
- Improved handling of single
gaQSAR objects versus lists
of objects.
- Updated permutation-test p-values with a plus-one correction.
- Improved double cross-validation predictor selection
frequencies.
- Updated examples, vignettes and documentation.
gaQSAR 1.2.1
Improvements
- Fix: permutation-test validation for
gaQSAR_dcv now
reconstructs outer predictions from saved fold models (removes spurious
validation warnings).
- Improved reproducibility:
bestSeed tracking and outer
fold seed are stored and used consistently.
- Clarified documentation and messages for validation and CV
bookkeeping.
gaQSAR 1.2.0
New Features
- Added nested cross-validation with outer CV (LOO/k-fold), inner
LOOCV fitness, robust error handling, AD thresholds, verbose output
option, and predictor stability metrics.
gaQSAR 1.0.1
New Features
- Added VIF (Variable Inflation Factor) scores
- Added counting of Williams plot objects with high residuals and
leverages for both training and test sets
- GA-selected models now report adjusted R-squared alongside training
R-squared
gaQSAR 1.0.0