Github user srowen commented on a diff in the pull request:
https://github.com/apache/spark/pull/3637#discussion_r21496412
--- 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 --
I'm also more sympathetic with a strongly-typed API here rather than
overload floating-point values to represent unordered categories. Are there
really so many possibilities? Any continuous or ordinal value really does
naturally translate to a double. Categoricals are the only other type of value
that needs a separate representation.
I feel like this misses some opportunities to optimize the internal
representation (e.g. a Dataset whose feature is known to be one of N values
doesn't need a double, but potentially just N bits) and avoid ambiguities of
representation (is negative -1? 0?) This is one of those areas where the
'simple' API just seems to push complexity elsewhere or ignore it. An algorithm
either has to have its own checks for whether 1.0 is a category or not, or,
overlooks the distinction. Same with the caller.
---
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
