Factor analysis

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Factor analysis

Jessica Koh
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

Diego Javier Zea
Hi Jessica

I didn't find a Julia implementation. It's mentioned in the future plans for MultivariateStats.jl and there is (not merged) PR in that package
It's implemented in scikit-learn, so you can used from Julia with: https://github.com/cstjean/ScikitLearn.jl

Best,

El sábado, 25 de junio de 2016, 19:18:27 (UTC-3), Jessica Koh escribió:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

colintbowers
In reply to this post by Jessica Koh
I haven't seen anything yet on traditional common factor analysis by maximum likelihood. Depending on your problem, you might be able to use principal components instead which is implemented in MultivariateStats.jl... e.g. in dual-asymptotic framework, simple transformations of the first k principal components are consistent estimators of the space-spanned by a k-dimensional common factor space.

Cheers,

Colin

On Sunday, 26 June 2016 08:18:27 UTC+10, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

Cedric St-Jean-2
You can also look into Madeleine Udell's LowRankModels.jl. It doesn't contain factor analysis, but unless I'm mistaken it should be possible to formulate it by specifying the objective function and regularizers appropriately

On Sunday, June 26, 2016 at 7:43:04 PM UTC-4, [hidden email] wrote:
I haven't seen anything yet on traditional common factor analysis by maximum likelihood. Depending on your problem, you might be able to use principal components instead which is implemented in MultivariateStats.jl... e.g. in dual-asymptotic framework, simple transformations of the first k principal components are consistent estimators of the space-spanned by a k-dimensional common factor space.

Cheers,

Colin

On Sunday, 26 June 2016 08:18:27 UTC+10, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

Alex Williams
In reply to this post by colintbowers

Hey Colin - could you send a link or reference to that? Sounds like something I'd like to read up on.

I'd really like to see a solid factor analysis implementation soon. As Diego said I think SciKitLearn.jl is the best stopgap option at the moment.

On Jun 26, 2016 4:43 PM, <[hidden email]> wrote:
I haven't seen anything yet on traditional common factor analysis by maximum likelihood. Depending on your problem, you might be able to use principal components instead which is implemented in MultivariateStats.jl... e.g. in dual-asymptotic framework, simple transformations of the first k principal components are consistent estimators of the space-spanned by a k-dimensional common factor space.

Cheers,

Colin

On Sunday, 26 June 2016 08:18:27 UTC+10, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

Alex Williams
In reply to this post by Cedric St-Jean-2

@Cedric - I don't think Madeleine's framework includes factor analysis at the moment. Particularly if there is missing data one would have to iteratively alternate between estimating the mean/variance of each feature and the factors.

On Jun 26, 2016 5:20 PM, "Cedric St-Jean" <[hidden email]> wrote:
You can also look into Madeleine Udell's LowRankModels.jl. It doesn't contain factor analysis, but unless I'm mistaken it should be possible to formulate it by specifying the objective function and regularizers appropriately

On Sunday, June 26, 2016 at 7:43:04 PM UTC-4, [hidden email] wrote:
I haven't seen anything yet on traditional common factor analysis by maximum likelihood. Depending on your problem, you might be able to use principal components instead which is implemented in MultivariateStats.jl... e.g. in dual-asymptotic framework, simple transformations of the first k principal components are consistent estimators of the space-spanned by a k-dimensional common factor space.

Cheers,

Colin

On Sunday, 26 June 2016 08:18:27 UTC+10, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

colintbowers
In reply to this post by Alex Williams
Sure, no problems. The gentlest introduction I know of (and it is still fairly heavy reading) is Bai, Ng (2006) "Evaluating Latent and Observed Factors in Macroeconomics and Finance" in the Journal of Econometrics. It contains references to all the really heavy theoretical papers too if you're interested. Probably also worth mentioning Bai, Ng (2002) "Determining the Number of Factors in an Approximate Factor Model" in Econometrica, as this material is necessary to consistently estimate the dimension of the common factor space.

If you don't have access to these journals, Serena Ng has pdfs and matlab code for both papers at her homepage here: http://www.columbia.edu/~sn2294/pub.html

At some point or other I implemented the techniques in both papers in matlab code too (Serena didn't have matlab code available at the time) so let me know if you want a copy (I didn't get round to posting it on file-exchange). If I had more free time I would probably have already made a Julia package of this stuff, but kids = no free time :-)

Cheers,

Colin

On Monday, 27 June 2016 10:22:46 UTC+10, Alex Williams wrote:

Hey Colin - could you send a link or reference to that? Sounds like something I'd like to read up on.

I'd really like to see a solid factor analysis implementation soon. As Diego said I think SciKitLearn.jl is the best stopgap option at the moment.

On Jun 26, 2016 4:43 PM, <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="dYF2zUg1AQAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">colint...@...> wrote:
I haven't seen anything yet on traditional common factor analysis by maximum likelihood. Depending on your problem, you might be able to use principal components instead which is implemented in MultivariateStats.jl... e.g. in dual-asymptotic framework, simple transformations of the first k principal components are consistent estimators of the space-spanned by a k-dimensional common factor space.

Cheers,

Colin

On Sunday, 26 June 2016 08:18:27 UTC+10, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="dYF2zUg1AQAJ" rel="nofollow" onmousedown="this.href=&#39;javascript:&#39;;return true;" onclick="this.href=&#39;javascript:&#39;;return true;">julia-stats...@googlegroups.com.
For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;" onclick="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;">https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

Alan Crawford
To those interested in factor models you may find Matthieu Gomez's SparseFactorModels.jl useful.

Best
Alan

On Monday, 27 June 2016 02:04:02 UTC+1, [hidden email] wrote:
Sure, no problems. The gentlest introduction I know of (and it is still fairly heavy reading) is Bai, Ng (2006) "Evaluating Latent and Observed Factors in Macroeconomics and Finance" in the Journal of Econometrics. It contains references to all the really heavy theoretical papers too if you're interested. Probably also worth mentioning Bai, Ng (2002) "Determining the Number of Factors in an Approximate Factor Model" in Econometrica, as this material is necessary to consistently estimate the dimension of the common factor space.

If you don't have access to these journals, Serena Ng has pdfs and matlab code for both papers at her homepage here: <a href="http://www.columbia.edu/~sn2294/pub.html" target="_blank" rel="nofollow" onmousedown="this.href=&#39;http://www.google.com/url?q\x3dhttp%3A%2F%2Fwww.columbia.edu%2F~sn2294%2Fpub.html\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNH92UlaQHPBeZ2ETwWZBFJkR9VFLg&#39;;return true;" onclick="this.href=&#39;http://www.google.com/url?q\x3dhttp%3A%2F%2Fwww.columbia.edu%2F~sn2294%2Fpub.html\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNH92UlaQHPBeZ2ETwWZBFJkR9VFLg&#39;;return true;">http://www.columbia.edu/~sn2294/pub.html

At some point or other I implemented the techniques in both papers in matlab code too (Serena didn't have matlab code available at the time) so let me know if you want a copy (I didn't get round to posting it on file-exchange). If I had more free time I would probably have already made a Julia package of this stuff, but kids = no free time :-)

Cheers,

Colin

On Monday, 27 June 2016 10:22:46 UTC+10, Alex Williams wrote:

Hey Colin - could you send a link or reference to that? Sounds like something I'd like to read up on.

I'd really like to see a solid factor analysis implementation soon. As Diego said I think SciKitLearn.jl is the best stopgap option at the moment.

On Jun 26, 2016 4:43 PM, <[hidden email]> wrote:
I haven't seen anything yet on traditional common factor analysis by maximum likelihood. Depending on your problem, you might be able to use principal components instead which is implemented in MultivariateStats.jl... e.g. in dual-asymptotic framework, simple transformations of the first k principal components are consistent estimators of the space-spanned by a k-dimensional common factor space.

Cheers,

Colin

On Sunday, 26 June 2016 08:18:27 UTC+10, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to julia-stats...@googlegroups.com.
For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;" onclick="this.href=&#39;https://groups.google.com/d/optout&#39;;return true;">https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Factor analysis

parthasarathy ganguly
In reply to this post by Jessica Koh
I found the easiest way to do factor analysis is to call r using RCall and enter $ sign. Then you can use r code and do factor analysis.

On Sunday, 26 June 2016 03:48:27 UTC+5:30, Jessica Koh wrote:
Hello,

Is factor analysis currently being developed?

--
You received this message because you are subscribed to the Google Groups "julia-stats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Loading...