-
Added five new scoring methods: correlation, random forest feature importance, information gain, area under the ROC curve, and cross tabulation filters.
-
Transitioned from S3 to S7 and refactored S7 for improved clarity, maintainability, and performance.
-
Added multiple new functions to support ranking of single score.
show_best_score_prop(),show_best_score_num(),show_best_score_cutoff(),show_best_score_dual(),rank_best_score_min()andrank_best_score_dense()can be used to select and rank single score. -
Added two new functions to support ranking of multiple scores.
show_best_desirability_prop()andshow_best_desirability_num()can be used for multiparameter optimization via desirability functions. -
Added multiple utility functions to support handling of both single and multiple scores:
arrange_score(),fill_safe_value(),bind_scores(), andfill_safe_values().