[
https://issues.apache.org/jira/browse/FLINK-1844?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14576282#comment-14576282
]
Faye Beligianni commented on FLINK-1844:
----------------------------------------
Hey [~tvas]
I opened a PR for the normaliser, which I named MinMaxScaler.
Any comments are welcomed!
Regarding the two tests that I wrote, I think that maybe they are too simple,
as I am only checking if the numbers are in the user-specified range.
An attempt to cross check the result against a dataset of
"expectedScaledVectors" would require to use the same method for calculating
the "expectedScaledVectors" which I used in the implementation of the
MinMaxScaler (wasn't sure if that would've been correct).
> Add Normaliser to ML library
> ----------------------------
>
> Key: FLINK-1844
> URL: https://issues.apache.org/jira/browse/FLINK-1844
> Project: Flink
> Issue Type: Improvement
> Components: Machine Learning Library
> Reporter: Faye Beligianni
> Assignee: Faye Beligianni
> Priority: Minor
> Labels: ML, Starter
>
> In many algorithms in ML, the features' values would be better to lie between
> a given range of values, usually in the range (0,1) [1]. Therefore, a
> {{Transformer}} could be implemented to achieve that normalisation.
> Resources:
> [1][http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
