Github user jkbradley commented on a diff in the pull request: https://github.com/apache/spark/pull/3637#discussion_r21559541 --- Diff: mllib/src/main/scala/org/apache/spark/ml/LabeledPoint.scala --- @@ -0,0 +1,52 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.spark.ml + +import scala.beans.BeanInfo + +import org.apache.spark.annotation.AlphaComponent +import org.apache.spark.mllib.linalg.Vector + +/** + * :: AlphaComponent :: + * Class that represents an instance (data point) for prediction tasks. + * + * @param label Label to predict + * @param features List of features describing this instance + * @param weight Instance weight + */ +@AlphaComponent +@BeanInfo +case class LabeledPoint(label: Double, features: Vector, weight: Double) { --- End diff -- The main pros & cons I see for having continuous & categorical types for labels & features are: Pros: * Type safety. Cons: * Algorithms may need to make copies of the data, depending on how much we expose internals of a features type. * Users may have to worry more about types. * For labels, if we load data from a file without metadata (like libsvm), we may need to assume that everything is continuous. Users will have to explicitly cast labels to categorical for classification. * For features, strong typing implies a stronger contract, where the assumption is that users specify the correct types. I've been wondering about having more "best effort" APIs, where we take suggestions from users (like DecisionTree's categoricalFeaturesInfo) but otherwise try to infer the best types to use under the hood. These lists started out much more balanced, but I guess I'm voting for the old system where everything is a Double.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org