[
https://issues.apache.org/jira/browse/FLINK-2259?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15243774#comment-15243774
]
ASF GitHub Bot commented on FLINK-2259:
---------------------------------------
GitHub user rawkintrevo opened a pull request:
https://github.com/apache/flink/pull/1898
[FLINK-2259][ml] Add Train-Testing Splitters
This PR adds an object in ml/pipeline called splitter with the following
methods:
randomSplit: Splits a DataSet into two data sets using DataSet.sample
multiRandomSplit: Splits a DataSet into multiple datasets according to an
array of proportions
kFoldSplit: Splits DataSet into k TrainTest objects which have a testing
data set of size 1/k of the original dataset and the remainder of the dataset
in the training
trainTestSplit: A wrapper for randomSplit that return a TrainTest object
trainTestHoldoutSplit: A wrapper for multiRandomSplit that returns a
TrainTestHoldout object
the TrainTest and TrainTestHoldout objects are case classes. randomSplit
and multiRandomSplit return arrays of DataSets.
- [x] General
- [ ] Documentation
- Documentation is in code, will write markdown after
review/feedback/finalization
- [x] Tests & Build
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/rawkintrevo/flink train-test-split
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/flink/pull/1898.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #1898
----
commit ec1e65a31d80b33589b73619f2a5dd0a8e09c568
Author: Trevor Grant <[email protected]>
Date: 2016-04-15T22:37:51Z
Add Splitter Pre-processing
commit 3ecdc3818dd11a847136510dabe96f444924d319
Author: Trevor Grant <[email protected]>
Date: 2016-04-15T22:40:38Z
Add Splitter Pre-processing
----
> Support training Estimators using a (train, validation, test) split of the
> available data
> -----------------------------------------------------------------------------------------
>
> Key: FLINK-2259
> URL: https://issues.apache.org/jira/browse/FLINK-2259
> Project: Flink
> Issue Type: New Feature
> Components: Machine Learning Library
> Reporter: Theodore Vasiloudis
> Assignee: Trevor Grant
> Priority: Minor
> Labels: ML
>
> When there is an abundance of data available, a good way to train models is
> to split the available data into 3 parts: Train, Validation and Test.
> We use the Train data to train the model, the Validation part is used to
> estimate the test error and select hyperparameters, and the Test is used to
> evaluate the performance of the model, and assess its generalization [1]
> This is a common approach when training Artificial Neural Networks, and a
> good strategy to choose in data-rich environments. Therefore we should have
> some support of this data-analysis process in our Estimators.
> [1] Friedman, Jerome, Trevor Hastie, and Robert Tibshirani. The elements of
> statistical learning. Vol. 1. Springer, Berlin: Springer series in
> statistics, 2001.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
