trees.regression — cart 0.1.0-SNAPSHOT

create-leaf clj

(create-leaf mean answer)

Given a mapping from class -> count in node, creates a tree leaf node

feature-type-predicate clj

(feature-type-predicate value feature-type)

find-best-splitting-feature clj

(find-best-splitting-feature data features target parent-impurity)

Searches across all features to find the one with lowest error

learn clj

(learn data features target)

(learn data features target min-count current-depth max_depth)

log-message clj

(log-message target-values current-depth)

predict clj

(predict tree x)

Given a fully constructed decision tree T and a record X, return the predicted label

pure? clj

(pure? response-counts)

Determines if only a single class is within a node