[ https://issues.apache.org/jira/browse/SPARK-2692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14074769#comment-14074769 ]
Joseph K. Bradley commented on SPARK-2692: ------------------------------------------ https://github.com/apache/spark/pull/1582 > Decision Tree API update > ------------------------ > > Key: SPARK-2692 > URL: https://issues.apache.org/jira/browse/SPARK-2692 > Project: Spark > Issue Type: Improvement > Components: MLlib > Reporter: Joseph K. Bradley > > Split Decision Tree API into separate Classifier and Regressor classes. > Details: > (a) Split classes: E.g.: DecisionTree --> DecisionTreeClassifier and > DecisionTreeRegressor > (b) Included print() function for human-readable model descriptions > (c) Renamed Strategy to *Params. Changed to take strings instead of special > types. > (d) Made configuration classes (Impurity, QuantileStrategy) private to mllib. > (e) Changed meaning of maxDepth by 1 to match scikit-learn and rpart. > (f) Removed static train() functions in favor of using Params classes. > (g) Introduced DatasetInfo class for metadata. -- This message was sent by Atlassian JIRA (v6.2#6252)