dtGAP 0.0.2
New Features
save_dtGAP(): Export dtGAP visualizations to PNG, PDF,
or SVG files with customizable dimensions and resolution.
select_vars parameter in dtGAP():
Display-only variable filtering for heatmap panels while the tree is
trained on all variables.
fit and user_var_imp parameters in
dtGAP(): Supply a pre-trained tree (rpart, party, or caret)
directly, with automatic model detection and optional user-provided
variable importance.
interactive parameter in dtGAP(): Launch a
Shiny-based interactive heatmap viewer via
InteractiveComplexHeatmap.
compare_dtGAP(): Compare two or more tree models
side-by-side on a single wide canvas.
- Random forest extension via
partykit::cforest:
train_rf(): Train a conditional random forest and
extract variable importance.
rf_summary(): Ensemble-level summary with variable
importance barplot and representative tree identification.
rf_dtGAP(): Visualize any individual tree from the
forest using the full dtGAP pipeline.
Bug Fixes
- Fix
formatC() error in prepare_tree() for
cforest trees that lack numeric p-values.
Documentation
- Updated vignette with usage examples for all new features.
- Updated README with new feature descriptions and code examples.
dtGAP 0.0.1
- Initial release.
- Core
dtGAP() function for supervised decision-tree
visualization using the GAP framework.
- Support for rpart, party, C5.0, and caret tree models.
- Confusion matrix maps, decision-tree matrix maps, predicted class
membership maps, and evaluation panels.
- Row and column proximity with seriation support.
- Classification and regression tasks.
- Seven built-in datasets: Psychosis_Disorder, penguins, wine,
diabetes, train_covid/test_covid, wine_quality_red, galaxy.