commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilles <gil...@harfang.homelinux.org>
Subject Re: MATH cKMeans Implementation
Date Thu, 09 Nov 2017 13:02:45 GMT
Hello.

On Thu, 9 Nov 2017 10:15:11 +0530, Shubham Jindal wrote:
> Hello,
> Thanks for the update Gilles. How should I proceed with the patch 
> request?
> Shall I go ahead and create a feature request on JIRA and then have a 
> patch
> request tied to that feature?

Yes, that is fine.
However, to minimize rounds of updates, be sure to
  * check out the git "master" branch,
  * create a "feature branch" (see file in "doc/development"),
  * run "mvn site" in the project directory,
  * ensure that all the reports (that will be located in the
    "target/site" directory) are clean, and
  * cover the provided code with unit tests.

Thanks for your interest in Commons Math.

Best regards,
Gilles

> Please let me know the procedure to go ahead
> with this
>
> Thanks
> Shubham Jindal
>
> On Thu, Nov 9, 2017 at 5:44 AM, Gilles <gilles@harfang.homelinux.org> 
> wrote:
>
>> Hi.
>>
>> On Wed, 8 Nov 2017 21:20:19 +0530, Shubham Jindal wrote:
>>
>>> Hello,
>>>
>>> I have written a full fledged efficient implementation of cKMeans 
>>> in Java
>>> https://urldefense.proofpoint.com/v2/url?u=https-3A__cran.r-
>>> 2Dproject.org_web_packages_Ckmeans.1d.dp_index.html&d=DwICaQ
>>> &c=3_5eq9zYXWRS8ywqccmjfTRKM8mRLoQp6HBg1Tdb_Pc&r=O38Sg7Ci5zX
>>> 0AvRwc1e-g-mbng-04QjfLbi_9z_-CTc&m=xukM8HSh3tlEGIT_QDHqyHxR
>>> 
>>> XotWOzZvFa7CPvzlInI&s=mjhfzhAKtcDskCCsKnVkHogxv7r31FKEaF8MpK9dnQo&e=

>>> and
>>>
>>> https://urldefense.proofpoint.com/v2/url?u=https-3A__journal
>>> .r-2Dproject.org_archive_2011-2D2_RJournal-5F2011-2D2-
>>> 5FWang-2BSong.pdf&d=DwICaQ&c=3_5eq9zYXWRS8ywqccmjfTRKM8mRLo
>>> Qp6HBg1Tdb_Pc&r=O38Sg7Ci5zX0AvRwc1e-g-mbng-04QjfLbi_9z_-CTc&
>>> m=xukM8HSh3tlEGIT_QDHqyHxRXotWOzZvFa7CPvzlInI&s=Qlw5usWlEii6
>>> 4hiLiFEF9nBN94x7K0td6PgILBpum08&e=
>>>
>>> The algorithm described here is *O(kn^2)* where *k*: number of 
>>> clusters
>>> and
>>> *n*: number of 1D points. But, there exists an efficient 
>>> implementation in
>>> later versions of cKMeans which is *O(knlog(n))*
>>>
>>> cKMeans is faster than kMeans and also deterministic in nature. Can 
>>> I
>>> submit a patch request for cKMeans implementation in Apache Commons 
>>> Math3
>>> ML Clustering
>>>
>>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__
>>> commons.apache.org_proper_commons-2Dmath_javadocs_api-2D3.6_
>>> org_apache_commons_math3_ml_clustering_package-2Dsummary.
>>> html&d=DwICaQ&c=3_5eq9zYXWRS8ywqccmjfTRKM8mRLoQp6HBg1Tdb_Pc&
>>> r=O38Sg7Ci5zX0AvRwc1e-g-mbng-04QjfLbi_9z_-CTc&m=xukM8HSh3tlE
>>> GIT_QDHqyHxRXotWOzZvFa7CPvzlInI&s=i31GoF_7zNdr1h979W-oFpIu8K
>>> zRvrpEt5L1bPfUrGU&e=>
>>> package
>>> as a contribution?
>>>
>>
>> Thanks for your proposal and interest in contributing.
>>
>> The current development branch (git "master") is towards
>> version 4.0 of the library:
>>   https://urldefense.proofpoint.com/v2/url?u=http-3A__commons.
>> apache.org_proper_commons-2Dmath_source-2Drepository.html&d=
>> DwICaQ&c=3_5eq9zYXWRS8ywqccmjfTRKM8mRLoQp6HBg1Tdb_Pc&r=O38Sg
>> 7Ci5zX0AvRwc1e-g-mbng-04QjfLbi_9z_-CTc&m=xukM8HSh3tlEGIT_QDH
>> 
>> qyHxRXotWOzZvFa7CPvzlInI&s=_0Vu9eCJBe8s_cNDKBBolT6vX09GA_V3_Qjd1uisM1o&e=
>>
>> Also, you should be aware that Commons Math has had very few
>> contributors for the past year.  A lot of work was done since
>> the last official release, but a lot is still needed in order
>> to be able to release the next one, due to the many open issues
>> and lack of human resources:
>>   https://urldefense.proofpoint.com/v2/url?u=https-3A__issues.
>> apache.org_jira_browse_MATH&d=DwICaQ&c=3_5eq9zYXWRS8ywqccmjf
>> TRKM8mRLoQp6HBg1Tdb_Pc&r=O38Sg7Ci5zX0AvRwc1e-g-mbng-04QjfLbi
>> _9z_-CTc&m=xukM8HSh3tlEGIT_QDHqyHxRXotWOzZvFa7CPvzlInI&s=S1G
>> 4GacI0e5qXM4w5yYLeqJFuL0Lg9JaDC-JLYv9hhg&e=
>>
>> It has been proposed to break the library into more manageable
>> standalone components (the contents of the "o.a.c.math4.ml"
>> package was among the likely candidates) but there was no
>> agreement within the Commons project management committee on
>> this attempt to revive development.[1]
>>
>> Best regards,
>> Gilles
>>
>> [1] Full story is in the "dev" ML archives:
>>        https://urldefense.proofpoint.com/v2/url?u=http-
>> 3A__markmail.org_list_org.apache.commons.dev_&d=DwICaQ&
>> c=3_5eq9zYXWRS8ywqccmjfTRKM8mRLoQp6HBg1Tdb_Pc&r=O38Sg7Ci5zX0
>> AvRwc1e-g-mbng-04QjfLbi_9z_-CTc&m=xukM8HSh3tlEGIT_QDHqyHxR
>> XotWOzZvFa7CPvzlInI&s=iJmDCZVfCOlxHCXIv8cIQjTw_TxMKgVuWYsguMk_xuk&e=
>>
>>
>> Thanks
>>> Shubham Jindal
>>>


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message