Module builder 

Tree building algorithms and utilities

This module contains algorithms for building decision trees, including split finding, impurity calculations, and feature grouping utilities.

Structs

BestFirstTreeBuilder

Best-first tree builder

Functions

apply_auto_correlation_grouping

Apply automatic correlation-based feature grouping

apply_feature_grouping

Apply feature grouping to training data

apply_hierarchical_grouping

Apply hierarchical clustering-based feature grouping

apply_manual_grouping

Apply manual feature grouping specified by user

calculate_correlation_matrix

Calculate correlation matrix for features

calculate_pearson_correlation

Calculate Pearson correlation between two feature vectors

create_reduced_feature_matrix

Create reduced feature matrix with only representative features

find_best_logloss_split

Find best split using Log-loss criterion for classification

find_best_mae_split

Find best split using MAE criterion for regression

find_best_split_for_node

Find best split for a node given sample indices

find_best_twoing_split

Find best split using Twoing criterion for classification

gini_impurity

Calculate gini impurity for multiway splits

handle_missing_values

Handle missing values in the data based on the specified strategy

hierarchical_clustering

Simple hierarchical clustering implementation

log_loss_impurity

Calculate Log-loss impurity for probability-based classification

mae_impurity

Calculate Mean Absolute Error (MAE) impurity for regression

select_group_representative

Select representative feature from a group

split_samples_by_threshold

Split samples by threshold

twoing_impurity

Calculate Twoing criterion impurity for binary classification