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()
and rank_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()
and
show_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()
, and
fill_safe_values()
.