[Statistics]Port codes from Commons Math

classic Classic list List threaded Threaded
46 messages Options
123
Reply | Threaded
Open this post in threaded view
|

Re: [Statistics] Port codes from Commons Math

Gilles Sadowski
Hi Gimhana.

On Fri, 18 May 2018 00:16:04 +0530, Gimhana Nadeeshan wrote:

> Hi all,
>
> We might want to create a public branch for that work in order to
>> merge PRs more quickly without risk of breaking "master".
>> What do you think?  Eric?
>>
>
> I ported the Statistics Interval Module and would like to get your
> reviews.
> How should I make the Pull request ?

I've just created a new branch on the repository; please make
all PR refer to "task_STATISTICS-5".
I also suggest that you create finer-grained "sub-tasks" of
   https://issues.apache.org/jira/browse/STATISTICS-5

Thanks,
Gilles

>
> Best Regards,
> Gimhana
>
>
> On 5 May 2018 at 18:50, Gilles <[hidden email]> wrote:
>
>> Hi Gimhana.
>>
>> On Sat, 5 May 2018 15:50:43 +0530, Gimhana Nadeeshan wrote:
>>
>>> Hello all,
>>>
>>> As I proposed early I would like to begin port code from
>>> Commons-math
>>> <https://github.com/apache/commons-math> to Commons-statistics
>>> <https://github.com/apache/commons-statistics>.
>>> (For further details refer my  GSoC Proposal
>>>
>>> <https://docs.google.com/document/d/1sqSa0hrYc2AD75RZyJRkeqC
>>> OBOqTOeMnPaBsE9U5YhU/edit?usp=sharing>
>>> though I'm not selected this year)
>>>
>>> This is my proposed architecture in brief
>>>
>>>    1. Commons-Statistics-Core => Frequency and StatUtils classes
>>> (Can add
>>>    more common classes while implementing)
>>>    2. Commons-Statistics-Correlation
>>>    3. Commons-Statistics-Descriptive
>>>    4. Commons-Statistics-Inference
>>>    5. Commons-Statistics-Interval
>>>    6. Commons-Statistics-Ranking
>>>    7. Commons-Statistics-Regression
>>>
>>
>> Nit-pick: module names have no capital in them (just a convention).
>> So: "commons-statistics-core" rather than "Commons-Statistics-Core",
>> etc.
>>
>> While I referring Commons-Geometry
>>>
>>
>> No need to refer to that project since "Commons Statistics" has been
>> set up:
>>   http://commons.apache.org/proper/commons-statistics/
>>
>> The code repository is here:
>>   https://git1-us-west.apache.org/repos/asf?p=commons-statisti
>> cs.git;a=tree
>> It already contains a "commons-statistics-distribution" module whose
>> layout can be duplicated in the modules which you are proposing
>> above
>> (with appropriate changes of course).
>>
>> ported code to get a head start , I
>>> found that each module inside, contain a pox.xml file. Are they
>>> implemented
>>> as separate projects and then group in the same package? I'm asking
>>> because
>>> Since I'm new to code porting :-).
>>>
>>
>> A requirement is that no package should be shared between different
>> modules; by convention, the top-level package of module
>>   commons-statistics-descriptive
>> would be
>>   org.apache.commons.statistics.descriptive
>>
>> [And so on for the other modules. But I'd suggest you start with
>> one.]
>>
>> If so in here should I create all 7 projects and then group those in
>> same
>>> project.
>>>
>>
>> No, the project is "Commons Statisitics" and it would contain
>> several
>> _maven_ modules, each of which should ultimately map to a _JPMS_
>> (JDK9)
>> module).
>>
>> Firstly I suppose to start port Ranking Module as it has less
>>> dependencies comparing to others.
>>>
>>
>> Fine. But don't forget to browse through the JIRA issues of Commons
>> Math (CM) for things that would need fixing.  Whenever it's the
>> case,
>> please open a report in the new JIRA project (linking to the CM
>> report), and post here your proposed solution (or questions).
>>
>> We might want to create a public branch for that work in order to
>> merge PRs more quickly without risk of breaking "master".
>> What do you think?  Eric?
>>
>> Would someone help me to get a head start ??
>>>
>>
>> What else do you need?
>>
>> Best regards,
>> Gilles
>>
>> Best Regards,
>>> Gimhana.
>>>
>>>
>>> [...]
>>>>>
>>>>



---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

[Numbers] More modules (Was: [Statistics] Port codes from Commons Math)

Gilles Sadowski
Hello Gimhana.

On Thu, 17 May 2018 23:32:15 +0200, Gilles wrote:

> Hi Gimhana.
>
> On Fri, 18 May 2018 00:16:04 +0530, Gimhana Nadeeshan wrote:
>> Hi all,
>>
>> We might want to create a public branch for that work in order to
>>> merge PRs more quickly without risk of breaking "master".
>>> What do you think?  Eric?
>>>
>>
>> I ported the Statistics Interval Module and would like to get your
>> reviews.
>> How should I make the Pull request ?

I've had a quick look, but as I've noted in the JIRA comment, revamping
stat codes from "Commons Math" into "Commons Statistics" is a lot of
work, part redesign, and part porting/adaptation.

You could more easily get acquainted with the formalities of
contributing
to "Commons" (e.g. conventions, reports, doc, etc.) by tackling tasks
where there is no redesign involved (except the simplification
concerning
e.g. the handling of exceptions).

Would you be interested in examining whether the codes which I
mentioned
in [1] could be ported to "Commons Numbers"?  These are smaller tasks
to
which I could devote some review time, while for "Commons Statistics" I
don't want to enter an open-ended discussion on how to improve the
design
without the involvment of users of the "Commons Math" stat codes.

Regards,
Gilles

[1] https://markmail.org/message/x6ng4ao62jjbj763

> I've just created a new branch on the repository; please make
> all PR refer to "task_STATISTICS-5".
> I also suggest that you create finer-grained "sub-tasks" of
>   https://issues.apache.org/jira/browse/STATISTICS-5
>
> Thanks,
> Gilles
>
>>
>> Best Regards,
>> Gimhana
>>
>>
>> On 5 May 2018 at 18:50, Gilles <[hidden email]> wrote:
>>
>>> Hi Gimhana.
>>>
>>> On Sat, 5 May 2018 15:50:43 +0530, Gimhana Nadeeshan wrote:
>>>
>>>> Hello all,
>>>>
>>>> As I proposed early I would like to begin port code from
>>>> Commons-math
>>>> <https://github.com/apache/commons-math> to Commons-statistics
>>>> <https://github.com/apache/commons-statistics>.
>>>> (For further details refer my  GSoC Proposal
>>>>
>>>> <https://docs.google.com/document/d/1sqSa0hrYc2AD75RZyJRkeqC
>>>> OBOqTOeMnPaBsE9U5YhU/edit?usp=sharing>
>>>> though I'm not selected this year)
>>>>
>>>> This is my proposed architecture in brief
>>>>
>>>>    1. Commons-Statistics-Core => Frequency and StatUtils classes
>>>> (Can add
>>>>    more common classes while implementing)
>>>>    2. Commons-Statistics-Correlation
>>>>    3. Commons-Statistics-Descriptive
>>>>    4. Commons-Statistics-Inference
>>>>    5. Commons-Statistics-Interval
>>>>    6. Commons-Statistics-Ranking
>>>>    7. Commons-Statistics-Regression
>>>>
>>>
>>> Nit-pick: module names have no capital in them (just a convention).
>>> So: "commons-statistics-core" rather than
>>> "Commons-Statistics-Core", etc.
>>>
>>> While I referring Commons-Geometry
>>>>
>>>
>>> No need to refer to that project since "Commons Statistics" has
>>> been
>>> set up:
>>>   http://commons.apache.org/proper/commons-statistics/
>>>
>>> The code repository is here:
>>>   https://git1-us-west.apache.org/repos/asf?p=commons-statisti
>>> cs.git;a=tree
>>> It already contains a "commons-statistics-distribution" module
>>> whose
>>> layout can be duplicated in the modules which you are proposing
>>> above
>>> (with appropriate changes of course).
>>>
>>> ported code to get a head start , I
>>>> found that each module inside, contain a pox.xml file. Are they
>>>> implemented
>>>> as separate projects and then group in the same package? I'm
>>>> asking
>>>> because
>>>> Since I'm new to code porting :-).
>>>>
>>>
>>> A requirement is that no package should be shared between different
>>> modules; by convention, the top-level package of module
>>>   commons-statistics-descriptive
>>> would be
>>>   org.apache.commons.statistics.descriptive
>>>
>>> [And so on for the other modules. But I'd suggest you start with
>>> one.]
>>>
>>> If so in here should I create all 7 projects and then group those
>>> in same
>>>> project.
>>>>
>>>
>>> No, the project is "Commons Statisitics" and it would contain
>>> several
>>> _maven_ modules, each of which should ultimately map to a _JPMS_
>>> (JDK9)
>>> module).
>>>
>>> Firstly I suppose to start port Ranking Module as it has less
>>>> dependencies comparing to others.
>>>>
>>>
>>> Fine. But don't forget to browse through the JIRA issues of Commons
>>> Math (CM) for things that would need fixing.  Whenever it's the
>>> case,
>>> please open a report in the new JIRA project (linking to the CM
>>> report), and post here your proposed solution (or questions).
>>>
>>> We might want to create a public branch for that work in order to
>>> merge PRs more quickly without risk of breaking "master".
>>> What do you think?  Eric?
>>>
>>> Would someone help me to get a head start ??
>>>>
>>>
>>> What else do you need?
>>>
>>> Best regards,
>>> Gilles
>>>
>>> Best Regards,
>>>> Gimhana.
>>>>
>>>>
>>>> [...]
>>>>>>
>>>>>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: [Statistics] Port codes from Commons Math

Gimhana Nadeeshan
In reply to this post by Gilles Sadowski
Hi all,

Porting
*Interval Module *

JIRA SUB-TASK : Interval Module Porting
<https://issues.apache.org/jira/browse/STATISTICS-6>

DESCRIPTION   : Porting and Redesign Module Architecture from Classes at
Commons.Math4.stat.Interval
<https://git-wip-us.apache.org/repos/asf?p=commons-math.git;a=tree;f=src/main/java/org/apache/commons/math4/stat/interval;h=02f2ecc45042b40458414b9bb2516968abdaaa52;hb=HEAD>
Basically I have Created a Core Module for containing all the common
classes(solving dependency issues). Separate Module for specialized classes.

STRUCTURE      :

commons-statistics-core
     --- Localizable.java
     --- LocalizedFormats.java

commons-statistics-interval
     --- AgrestiCoullInterval.java
     --- BinomialConfidenceInterval.java
     --- ClopperPearsonInterval.java
     --- ConfidenceInterval.java
     --- IntervalUtils.java
     --- NormalApproximationInterval.java
     --- WilsonScoreInterval.java

All the Exceptions are Replaced using IllegalArgumentException.

Reviews and open for Discussion.

Best Regards,
Gimhana.

On 18 May 2018 at 03:02, Gilles <[hidden email]> wrote:

> Hi Gimhana.
>
> On Fri, 18 May 2018 00:16:04 +0530, Gimhana Nadeeshan wrote:
>
>> Hi all,
>>
>> We might want to create a public branch for that work in order to
>>
>>> merge PRs more quickly without risk of breaking "master".
>>> What do you think?  Eric?
>>>
>>>
>> I ported the Statistics Interval Module and would like to get your
>> reviews.
>> How should I make the Pull request ?
>>
>
> I've just created a new branch on the repository; please make
> all PR refer to "task_STATISTICS-5".
> I also suggest that you create finer-grained "sub-tasks" of
>   https://issues.apache.org/jira/browse/STATISTICS-5
>
> Thanks,
> Gilles
>
>
>
>> Best Regards,
>> Gimhana
>>
>>
>> On 5 May 2018 at 18:50, Gilles <[hidden email]> wrote:
>>
>> Hi Gimhana.
>>>
>>> On Sat, 5 May 2018 15:50:43 +0530, Gimhana Nadeeshan wrote:
>>>
>>> Hello all,
>>>>
>>>> As I proposed early I would like to begin port code from Commons-math
>>>> <https://github.com/apache/commons-math> to Commons-statistics
>>>> <https://github.com/apache/commons-statistics>.
>>>> (For further details refer my  GSoC Proposal
>>>>
>>>> <https://docs.google.com/document/d/1sqSa0hrYc2AD75RZyJRkeqC
>>>> OBOqTOeMnPaBsE9U5YhU/edit?usp=sharing>
>>>> though I'm not selected this year)
>>>>
>>>> This is my proposed architecture in brief
>>>>
>>>>    1. Commons-Statistics-Core => Frequency and StatUtils classes (Can
>>>> add
>>>>    more common classes while implementing)
>>>>    2. Commons-Statistics-Correlation
>>>>    3. Commons-Statistics-Descriptive
>>>>    4. Commons-Statistics-Inference
>>>>    5. Commons-Statistics-Interval
>>>>    6. Commons-Statistics-Ranking
>>>>    7. Commons-Statistics-Regression
>>>>
>>>>
>>> Nit-pick: module names have no capital in them (just a convention).
>>> So: "commons-statistics-core" rather than "Commons-Statistics-Core", etc.
>>>
>>> While I referring Commons-Geometry
>>>
>>>>
>>>>
>>> No need to refer to that project since "Commons Statistics" has been
>>> set up:
>>>   http://commons.apache.org/proper/commons-statistics/
>>>
>>> The code repository is here:
>>>   https://git1-us-west.apache.org/repos/asf?p=commons-statisti
>>> cs.git;a=tree
>>> It already contains a "commons-statistics-distribution" module whose
>>> layout can be duplicated in the modules which you are proposing above
>>> (with appropriate changes of course).
>>>
>>> ported code to get a head start , I
>>>
>>>> found that each module inside, contain a pox.xml file. Are they
>>>> implemented
>>>> as separate projects and then group in the same package? I'm asking
>>>> because
>>>> Since I'm new to code porting :-).
>>>>
>>>>
>>> A requirement is that no package should be shared between different
>>> modules; by convention, the top-level package of module
>>>   commons-statistics-descriptive
>>> would be
>>>   org.apache.commons.statistics.descriptive
>>>
>>> [And so on for the other modules. But I'd suggest you start with one.]
>>>
>>> If so in here should I create all 7 projects and then group those in same
>>>
>>>> project.
>>>>
>>>>
>>> No, the project is "Commons Statisitics" and it would contain several
>>> _maven_ modules, each of which should ultimately map to a _JPMS_ (JDK9)
>>> module).
>>>
>>> Firstly I suppose to start port Ranking Module as it has less
>>>
>>>> dependencies comparing to others.
>>>>
>>>>
>>> Fine. But don't forget to browse through the JIRA issues of Commons
>>> Math (CM) for things that would need fixing.  Whenever it's the case,
>>> please open a report in the new JIRA project (linking to the CM
>>> report), and post here your proposed solution (or questions).
>>>
>>> We might want to create a public branch for that work in order to
>>> merge PRs more quickly without risk of breaking "master".
>>> What do you think?  Eric?
>>>
>>> Would someone help me to get a head start ??
>>>
>>>>
>>>>
>>> What else do you need?
>>>
>>> Best regards,
>>> Gilles
>>>
>>> Best Regards,
>>>
>>>> Gimhana.
>>>>
>>>>
>>>> [...]
>>>>
>>>>>
>>>>>>
>>>>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>
Reply | Threaded
Open this post in threaded view
|

Re: [Numbers] More modules (Was: [Statistics] Port codes from Commons Math)

Gimhana Nadeeshan
In reply to this post by Gilles Sadowski
Hi Gilles,

[1] https://markmail.org/message/x6ng4ao62jjbj763
>

Ok sure I'll try those out and ask here when I need support.


while for "Commons Statistics" I
> don't want to enter an open-ended discussion on how to improve the design
> without the involvment of users of the "Commons Math" stat codes.
>

Should I postpone till the other devs' active involvement ?

Best Regards,
Gimhana.

On 19 May 2018 at 22:02, Gilles <[hidden email]> wrote:

> Hello Gimhana.
>
> On Thu, 17 May 2018 23:32:15 +0200, Gilles wrote:
>
>> Hi Gimhana.
>>
>> On Fri, 18 May 2018 00:16:04 +0530, Gimhana Nadeeshan wrote:
>>
>>> Hi all,
>>>
>>> We might want to create a public branch for that work in order to
>>>
>>>> merge PRs more quickly without risk of breaking "master".
>>>> What do you think?  Eric?
>>>>
>>>>
>>> I ported the Statistics Interval Module and would like to get your
>>> reviews.
>>> How should I make the Pull request ?
>>>
>>
> I've had a quick look, but as I've noted in the JIRA comment, revamping
> stat codes from "Commons Math" into "Commons Statistics" is a lot of
> work, part redesign, and part porting/adaptation.
>
> You could more easily get acquainted with the formalities of contributing
> to "Commons" (e.g. conventions, reports, doc, etc.) by tackling tasks
> where there is no redesign involved (except the simplification concerning
> e.g. the handling of exceptions).
>
> Would you be interested in examining whether the codes which I mentioned
> in [1] could be ported to "Commons Numbers"?  These are smaller tasks to
> which I could devote some review time, while for "Commons Statistics" I
> don't want to enter an open-ended discussion on how to improve the design
> without the involvment of users of the "Commons Math" stat codes.
>
> Regards,
> Gilles
>
> [1] https://markmail.org/message/x6ng4ao62jjbj763
>
> I've just created a new branch on the repository; please make
>> all PR refer to "task_STATISTICS-5".
>> I also suggest that you create finer-grained "sub-tasks" of
>>   https://issues.apache.org/jira/browse/STATISTICS-5
>>
>> Thanks,
>> Gilles
>>
>>
>>> Best Regards,
>>> Gimhana
>>>
>>>
>>> On 5 May 2018 at 18:50, Gilles <[hidden email]> wrote:
>>>
>>> Hi Gimhana.
>>>>
>>>> On Sat, 5 May 2018 15:50:43 +0530, Gimhana Nadeeshan wrote:
>>>>
>>>> Hello all,
>>>>>
>>>>> As I proposed early I would like to begin port code from Commons-math
>>>>> <https://github.com/apache/commons-math> to Commons-statistics
>>>>> <https://github.com/apache/commons-statistics>.
>>>>> (For further details refer my  GSoC Proposal
>>>>>
>>>>> <https://docs.google.com/document/d/1sqSa0hrYc2AD75RZyJRkeqC
>>>>> OBOqTOeMnPaBsE9U5YhU/edit?usp=sharing>
>>>>> though I'm not selected this year)
>>>>>
>>>>> This is my proposed architecture in brief
>>>>>
>>>>>    1. Commons-Statistics-Core => Frequency and StatUtils classes (Can
>>>>> add
>>>>>    more common classes while implementing)
>>>>>    2. Commons-Statistics-Correlation
>>>>>    3. Commons-Statistics-Descriptive
>>>>>    4. Commons-Statistics-Inference
>>>>>    5. Commons-Statistics-Interval
>>>>>    6. Commons-Statistics-Ranking
>>>>>    7. Commons-Statistics-Regression
>>>>>
>>>>>
>>>> Nit-pick: module names have no capital in them (just a convention).
>>>> So: "commons-statistics-core" rather than "Commons-Statistics-Core",
>>>> etc.
>>>>
>>>> While I referring Commons-Geometry
>>>>
>>>>>
>>>>>
>>>> No need to refer to that project since "Commons Statistics" has been
>>>> set up:
>>>>   http://commons.apache.org/proper/commons-statistics/
>>>>
>>>> The code repository is here:
>>>>   https://git1-us-west.apache.org/repos/asf?p=commons-statisti
>>>> cs.git;a=tree
>>>> It already contains a "commons-statistics-distribution" module whose
>>>> layout can be duplicated in the modules which you are proposing above
>>>> (with appropriate changes of course).
>>>>
>>>> ported code to get a head start , I
>>>>
>>>>> found that each module inside, contain a pox.xml file. Are they
>>>>> implemented
>>>>> as separate projects and then group in the same package? I'm asking
>>>>> because
>>>>> Since I'm new to code porting :-).
>>>>>
>>>>>
>>>> A requirement is that no package should be shared between different
>>>> modules; by convention, the top-level package of module
>>>>   commons-statistics-descriptive
>>>> would be
>>>>   org.apache.commons.statistics.descriptive
>>>>
>>>> [And so on for the other modules. But I'd suggest you start with one.]
>>>>
>>>> If so in here should I create all 7 projects and then group those in
>>>> same
>>>>
>>>>> project.
>>>>>
>>>>>
>>>> No, the project is "Commons Statisitics" and it would contain several
>>>> _maven_ modules, each of which should ultimately map to a _JPMS_ (JDK9)
>>>> module).
>>>>
>>>> Firstly I suppose to start port Ranking Module as it has less
>>>>
>>>>> dependencies comparing to others.
>>>>>
>>>>>
>>>> Fine. But don't forget to browse through the JIRA issues of Commons
>>>> Math (CM) for things that would need fixing.  Whenever it's the case,
>>>> please open a report in the new JIRA project (linking to the CM
>>>> report), and post here your proposed solution (or questions).
>>>>
>>>> We might want to create a public branch for that work in order to
>>>> merge PRs more quickly without risk of breaking "master".
>>>> What do you think?  Eric?
>>>>
>>>> Would someone help me to get a head start ??
>>>>
>>>>>
>>>>>
>>>> What else do you need?
>>>>
>>>> Best regards,
>>>> Gilles
>>>>
>>>> Best Regards,
>>>>
>>>>> Gimhana.
>>>>>
>>>>>
>>>>> [...]
>>>>>
>>>>>>
>>>>>>>
>>>>>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [hidden email]
> For additional commands, e-mail: [hidden email]
>
>
Reply | Threaded
Open this post in threaded view
|

Re: [Numbers] More modules (Was: [Statistics] Port codes from Commons Math)

Gimhana Nadeeshan
Hello ,

While Porting  the codes in [1] I discovered below problems.

commons-numbers-rootsolver (with codes from package
> "org.apache.commons.math4.analysis.solver")
>

This needs some dependencies in the Module *commons.Math4.differentiation*.
But I couldn't find corresponding redesign in the commons.Numbers.

And Regarding simplifying Exceptions thrown, I suppose to implement a
separate class as commons-statistics-distribution-DistributionException
<https://github.com/apache/commons-statistics/blob/master/commons-statistics-distribution/src/main/java/org/apache/commons/statistics/distribution/DistributionException.java>

Open for Suggestions.

Best Regards,
Gimhana

[1] https://markmail.org/message/x6ng4ao62jjbj763

On 20 May 2018 at 16:07, Gimhana Nadeeshan <[hidden email]>
wrote:

> Hi Gilles,
>
> [1] https://markmail.org/message/x6ng4ao62jjbj763
>>
>
> Ok sure I'll try those out and ask here when I need support.
>
>
> while for "Commons Statistics" I
>> don't want to enter an open-ended discussion on how to improve the design
>> without the involvment of users of the "Commons Math" stat codes.
>>
>
> Should I postpone till the other devs' active involvement ?
>
> Best Regards,
> Gimhana.
>
> On 19 May 2018 at 22:02, Gilles <[hidden email]> wrote:
>
>> Hello Gimhana.
>>
>> On Thu, 17 May 2018 23:32:15 +0200, Gilles wrote:
>>
>>> Hi Gimhana.
>>>
>>> On Fri, 18 May 2018 00:16:04 +0530, Gimhana Nadeeshan wrote:
>>>
>>>> Hi all,
>>>>
>>>> We might want to create a public branch for that work in order to
>>>>
>>>>> merge PRs more quickly without risk of breaking "master".
>>>>> What do you think?  Eric?
>>>>>
>>>>>
>>>> I ported the Statistics Interval Module and would like to get your
>>>> reviews.
>>>> How should I make the Pull request ?
>>>>
>>>
>> I've had a quick look, but as I've noted in the JIRA comment, revamping
>> stat codes from "Commons Math" into "Commons Statistics" is a lot of
>> work, part redesign, and part porting/adaptation.
>>
>> You could more easily get acquainted with the formalities of contributing
>> to "Commons" (e.g. conventions, reports, doc, etc.) by tackling tasks
>> where there is no redesign involved (except the simplification concerning
>> e.g. the handling of exceptions).
>>
>> Would you be interested in examining whether the codes which I mentioned
>> in [1] could be ported to "Commons Numbers"?  These are smaller tasks to
>> which I could devote some review time, while for "Commons Statistics" I
>> don't want to enter an open-ended discussion on how to improve the design
>> without the involvment of users of the "Commons Math" stat codes.
>>
>> Regards,
>> Gilles
>>
>> [1] https://markmail.org/message/x6ng4ao62jjbj763
>>
>> I've just created a new branch on the repository; please make
>>> all PR refer to "task_STATISTICS-5".
>>> I also suggest that you create finer-grained "sub-tasks" of
>>>   https://issues.apache.org/jira/browse/STATISTICS-5
>>>
>>> Thanks,
>>> Gilles
>>>
>>>
>>>> Best Regards,
>>>> Gimhana
>>>>
>>>>
>>>> On 5 May 2018 at 18:50, Gilles <[hidden email]> wrote:
>>>>
>>>> Hi Gimhana.
>>>>>
>>>>> On Sat, 5 May 2018 15:50:43 +0530, Gimhana Nadeeshan wrote:
>>>>>
>>>>> Hello all,
>>>>>>
>>>>>> As I proposed early I would like to begin port code from Commons-math
>>>>>> <https://github.com/apache/commons-math> to Commons-statistics
>>>>>> <https://github.com/apache/commons-statistics>.
>>>>>> (For further details refer my  GSoC Proposal
>>>>>>
>>>>>> <https://docs.google.com/document/d/1sqSa0hrYc2AD75RZyJRkeqC
>>>>>> OBOqTOeMnPaBsE9U5YhU/edit?usp=sharing>
>>>>>> though I'm not selected this year)
>>>>>>
>>>>>> This is my proposed architecture in brief
>>>>>>
>>>>>>    1. Commons-Statistics-Core => Frequency and StatUtils classes (Can
>>>>>> add
>>>>>>    more common classes while implementing)
>>>>>>    2. Commons-Statistics-Correlation
>>>>>>    3. Commons-Statistics-Descriptive
>>>>>>    4. Commons-Statistics-Inference
>>>>>>    5. Commons-Statistics-Interval
>>>>>>    6. Commons-Statistics-Ranking
>>>>>>    7. Commons-Statistics-Regression
>>>>>>
>>>>>>
>>>>> Nit-pick: module names have no capital in them (just a convention).
>>>>> So: "commons-statistics-core" rather than "Commons-Statistics-Core",
>>>>> etc.
>>>>>
>>>>> While I referring Commons-Geometry
>>>>>
>>>>>>
>>>>>>
>>>>> No need to refer to that project since "Commons Statistics" has been
>>>>> set up:
>>>>>   http://commons.apache.org/proper/commons-statistics/
>>>>>
>>>>> The code repository is here:
>>>>>   https://git1-us-west.apache.org/repos/asf?p=commons-statisti
>>>>> cs.git;a=tree
>>>>> It already contains a "commons-statistics-distribution" module whose
>>>>> layout can be duplicated in the modules which you are proposing above
>>>>> (with appropriate changes of course).
>>>>>
>>>>> ported code to get a head start , I
>>>>>
>>>>>> found that each module inside, contain a pox.xml file. Are they
>>>>>> implemented
>>>>>> as separate projects and then group in the same package? I'm asking
>>>>>> because
>>>>>> Since I'm new to code porting :-).
>>>>>>
>>>>>>
>>>>> A requirement is that no package should be shared between different
>>>>> modules; by convention, the top-level package of module
>>>>>   commons-statistics-descriptive
>>>>> would be
>>>>>   org.apache.commons.statistics.descriptive
>>>>>
>>>>> [And so on for the other modules. But I'd suggest you start with one.]
>>>>>
>>>>> If so in here should I create all 7 projects and then group those in
>>>>> same
>>>>>
>>>>>> project.
>>>>>>
>>>>>>
>>>>> No, the project is "Commons Statisitics" and it would contain several
>>>>> _maven_ modules, each of which should ultimately map to a _JPMS_ (JDK9)
>>>>> module).
>>>>>
>>>>> Firstly I suppose to start port Ranking Module as it has less
>>>>>
>>>>>> dependencies comparing to others.
>>>>>>
>>>>>>
>>>>> Fine. But don't forget to browse through the JIRA issues of Commons
>>>>> Math (CM) for things that would need fixing.  Whenever it's the case,
>>>>> please open a report in the new JIRA project (linking to the CM
>>>>> report), and post here your proposed solution (or questions).
>>>>>
>>>>> We might want to create a public branch for that work in order to
>>>>> merge PRs more quickly without risk of breaking "master".
>>>>> What do you think?  Eric?
>>>>>
>>>>> Would someone help me to get a head start ??
>>>>>
>>>>>>
>>>>>>
>>>>> What else do you need?
>>>>>
>>>>> Best regards,
>>>>> Gilles
>>>>>
>>>>> Best Regards,
>>>>>
>>>>>> Gimhana.
>>>>>>
>>>>>>
>>>>>> [...]
>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [hidden email]
>> For additional commands, e-mail: [hidden email]
>>
>>
>
Reply | Threaded
Open this post in threaded view
|

[Numbers] Porting root solvers (Was: [Numbers] More modules [...])

Gilles Sadowski
Hi.

On Tue, 22 May 2018 08:40:35 +0530, Gimhana Nadeeshan wrote:

> Hello ,
>
> While Porting  the codes in [1] I discovered below problems.
>
> commons-numbers-rootsolver (with codes from package
>> "org.apache.commons.math4.analysis.solver")
>>
>
> This needs some dependencies in the Module
> *commons.Math4.differentiation*.
> But I couldn't find corresponding redesign in the commons.Numbers.

It has been put on the queue:
   https://issues.apache.org/jira/projects/NUMBERS/issues/NUMBERS-69
but someone has to be willing to do the suggested work,
and stay around for support and maintenance...

But this dependency isn't required for porting most of the
solvers.  The "DerivativeStructure"-related code is only needed
for the current implementation of "NewtonRaphsonSolver" which
could be left behind until issue NUMBERS-69 is handled.

Regards,
Gilles

> [...]
>>> You could more easily get acquainted with the formalities of
>>> contributing
>>> to "Commons" (e.g. conventions, reports, doc, etc.) by tackling
>>> tasks
>>> where there is no redesign involved (except the simplification
>>> concerning
>>> e.g. the handling of exceptions).
>>>
>>> Would you be interested in examining whether the codes which I
>>> mentioned
>>> in [1] could be ported to "Commons Numbers"?  These are smaller
>>> tasks to
>>> which I could devote some review time, while for "Commons
>>> Statistics" I
>>> don't want to enter an open-ended discussion on how to improve the
>>> design
>>> without the involvment of users of the "Commons Math" stat codes.
>>>
>>> Regards,
>>> Gilles
>>>
>>> [1] https://markmail.org/message/x6ng4ao62jjbj763
>>>
>>>> [...]


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

123