spark-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From freeman-lab <...@git.apache.org>
Subject [GitHub] spark pull request: Streaming mllib [SPARK-2438][MLLIB]
Date Fri, 01 Aug 2014 20:51:57 GMT
Github user freeman-lab commented on a diff in the pull request:

    https://github.com/apache/spark/pull/1361#discussion_r15718348
  
    --- Diff: mllib/src/main/scala/org/apache/spark/mllib/regression/StreamingLinearAlgorithm.scala
---
    @@ -0,0 +1,83 @@
    +/*
    + * 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.mllib.regression
    +
    +import org.apache.spark.annotation.DeveloperApi
    +import org.apache.spark.Logging
    +import org.apache.spark.streaming.dstream.DStream
    +
    +/**
    + * :: DeveloperApi ::
    --- End diff --
    
    I've been testing this and it seems fairly robust, agreed we should clarify in the documentation.

    
    What I've tried:
    
    - If train and predict are called on the same stream (or on two streams with data arriving
simultaneously), order matters. If trainOn is first, the prediction will always use the subsequently
updated model. If predictOn is first, it will use the model from the previous update. In practice,
over multiple updates, either behavior seems reasonable, but maybe there should be a helpful
warning if the user calls predictOn before trainOn?
    
    - If they are called on different streams and the data arrive sequentially, order doesn't
matter. For example, if data arrive in the predictOn stream before the trainOn stream, the
prediction uses the intial weights (as it should) to predict, regardless of the order of the
calls.
    
    - It's ok, and maybe useful, to call predictOn repeatedly on different streams. For example,
training on one stream, and predicting on it and another, behaves correctly (modolu the ordering
issues described above).
    
    - If you call trainOn repeatedly on different streams, it will do an update when data
arrive in either stream, which seems fine. Could be used to update using multiple input sources.


---
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 infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

Mime
View raw message