You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "Mauro Talevi (JIRA)" <ji...@apache.org> on 2008/04/28 00:56:55 UTC

[jira] Created: (MATH-203) Add general multiple linear regression

Add general multiple linear regression 
---------------------------------------

                 Key: MATH-203
                 URL: https://issues.apache.org/jira/browse/MATH-203
             Project: Commons Math
          Issue Type: New Feature
            Reporter: Mauro Talevi
             Fix For: 2.0


Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 

The regression component has no external dependency outside of commons-math.

Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 

If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.



-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (MATH-203) Add general multiple linear regression

Posted by "Luc Maisonobe (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12595917#action_12595917 ] 

Luc Maisonobe commented on MATH-203:
------------------------------------

I finally found some time to review this patch.

The only thing that really bother me is the lack of error handling. The calculateBeta method could fail if the data is insufficient (say no data at all, or less data than problem dimension, or duplicated data leading to too few independent points).

Concerning JDK 1.5, I'll start a specific thread on the dev list to talk about this.

> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Issue Comment Edited: (MATH-203) Add general multiple linear regression

Posted by "Mauro Talevi (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12596019#action_12596019 ] 

maurotalevi edited comment on MATH-203 at 5/12/08 7:51 AM:
------------------------------------------------------------

ATM invalid or insufficient data would be caught by the RealMatrix handling and thrown as IllegalArgumentExceptions.  For sure input validation can be improved and more explicit runtime exceptions can be thrown.  



      was (Author: maurotalevi):
    ATM, invalid or insufficient data would be caught by the RealMatrix handling and throw IllegalArgumentExceptions.  Sure - better input validation and more explicit runtime exceptions can be thrown.  


  
> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (MATH-203) Add general multiple linear regression

Posted by "Luc Maisonobe (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12597815#action_12597815 ] 

Luc Maisonobe commented on MATH-203:
------------------------------------

The patch has been committed as of r657570 on 2.0 branch.
The following minor changes have been made before commit:
  - some missing '.' characters at the end of javadoc headers have been added
  - @authors tags have been removed (authors are cited in the dedicated section
    of the pom file, hence allowing maven to link to them in the generated site)
  - changes to compile with Java 5 have been committed previously

I'm not closing this issue yet as a few things are still missing:
 - enhanced error handling
 - documentation (in package.html and in the user manual)

> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Updated: (MATH-203) Add general multiple linear regression

Posted by "Mauro Talevi (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mauro Talevi updated MATH-203:
------------------------------

    Attachment: error-handling-and-docs-patch.txt

Attaching patch for improved error handling upon data input and update of docs stats page.


> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: error-handling-and-docs-patch.txt, multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Resolved: (MATH-203) Add general multiple linear regression

Posted by "Luc Maisonobe (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Luc Maisonobe resolved MATH-203.
--------------------------------

    Resolution: Fixed

fixed as of r658645 in branch 2.0

> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: error-handling-and-docs-patch.txt, multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (MATH-203) Add general multiple linear regression

Posted by "Mauro Talevi (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12596019#action_12596019 ] 

Mauro Talevi commented on MATH-203:
-----------------------------------

ATM, invalid or insufficient data would be caught by the RealMatrix handling and throw IllegalArgumentExceptions.  Sure - better input validation and more explicit runtime exceptions can be thrown.  



> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (MATH-203) Add general multiple linear regression

Posted by "Brent Worden (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12593048#action_12593048 ] 

Brent Worden commented on MATH-203:
-----------------------------------

Support for 1.4 should continue for at least until J2SE 1.4 reaches its end of life which Sun has slated for October of this year.  Even after that time, support for 1.4 should continue for a good deal longer.  We've maintained 1.3 support and that reached its end of life in 2006.

> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Updated: (MATH-203) Add general multiple linear regression

Posted by "Mauro Talevi (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mauro Talevi updated MATH-203:
------------------------------

    Attachment: multiple-linear-regression-patch.txt

Attaching initial patch for multiple linear regression, based on MATH_2_0 branch.

Note that the code relies on JDK 1.5 and as such the pom.xml has been updated to use compiler source/target=1.5 and junit 4.4.  

The code could be refactored down to JDK 1.4 although I think the 2.0 branch is an excellent opportunity to upgrade to use JDK 1.5 features such as annotations/generics/etc ...




> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (MATH-203) Add general multiple linear regression

Posted by "Mauro Talevi (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/MATH-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12593063#action_12593063 ] 

Mauro Talevi commented on MATH-203:
-----------------------------------

Support for 1.4 could certainly continue in 1.x branch, and the regression code could be also backported to the 1.x branch.  IMO the move to 1.5 as a base JDK has much stronger motivation that just the Sun end of life, given the amount of new language features.  Several OS projects have JDK 1.3 or 1.4 compat for 1.x releases and JDK 1.5 for 2.x. 

  


> Add general multiple linear regression 
> ---------------------------------------
>
>                 Key: MATH-203
>                 URL: https://issues.apache.org/jira/browse/MATH-203
>             Project: Commons Math
>          Issue Type: New Feature
>            Reporter: Mauro Talevi
>             Fix For: 2.0
>
>         Attachments: multiple-linear-regression-patch.txt
>
>
> Multiple regression is a fundamental element of several mathematical and statistical projects, among them econometrics one.   A general linear multiple regression is contained in the JET Regression component at http://jet.codehaus.org/javadoc/jet-regression, but it could benefit a wider community if it was part of commons-math. 
> The regression component has no external dependency outside of commons-math.
> Code for OLS and GLS multiple linear regression can be checked out from https://svn.codehaus.org/jet/trunk/jet-regression. 
> If this is of interest, I'll re-package the code to live in org.apache.commons.math.stat.regression.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.