hnp_umbrella() has a new interface. It now takes
separate X and Y arguments together with
importance_order instead of a single data frame
S and a class_col. It supports an arbitrary
number of ordered classes (T >= 2) rather than only
ternary classification.hnp_summary() now takes classifier,
X, Y and importance_order instead
of data and class_col, and accepts classifiers
that return class labels, probability matrices or score matrices, as
well as fitted model objects.hnp_map_classes() now accepts a variable number of
class labels via ... (in decreasing priority order) instead
of the fixed class_1, class_2,
class_3 arguments.probability_to_score_1(),
probability_to_score_2(), hnp_umbrella_flex()
and hnp_box_plot().T >= 2).hnp_umbrella() via the
pretrained_model and input_is_score
arguments.grid_search, grid_set, max_grid,
max_combinations) to minimize the weighted
misclassification objective, with a recursive multi-class threshold
search.hnp_boxplot() to visualize and summarize
under-classification and overall error from confusion matrices,
supporting single- and two-method comparisons.gen_data(), gen_normal_data(),
generate_ball_data(), etc.) and a neural-network scoring
helper (train_nn_and_get_scores()).hnp_delta_search() now uses
stats::pbinom() instead of an explicit
choose()-based summation, avoiding numerical overflow and
underflow for large samples.hnp_upper_bound() is now vectorized over the score
functions and adds validation for non-finite, NA and
length-mismatched score outputs.
Need a high-speed mirror for your open-source project?
Contact our mirror admin team at info@clientvps.com.
This archive is provided as a free public service to the community.
Proudly supported by infrastructure from VPSPulse , RxServers , BuyNumber , UnitVPS , OffshoreName and secure payment technology by ArionPay.