Account Options

  1. Sign in
The old Google Groups will be going away soon, but your browser is incompatible with the new version.
Google Groups Home
« Groups Home
Group info
Active older topics
20 new of 20 - May 16
7 new of 7 - May 14
19 new of 19 - May 14
4 new of 4 - May 13
8 new of 8 - May 12
8 new of 8 - May 11
1 new of 1 - May 10
3 new of 3 - May 10
18 new of 18 - May 9
3 new of 3 - May 9
Discussions
View:  Topic list, Topic summary Topics 1 - 10 of 1655  Older »

You cannot post messages because only members can post, and you are not currently a member.
Description: We are discussing the implementation of statistical and econometrics models and methods of data handling in Python.

The initial focus is on scikits.statsmodels, http://statsmodels.sourceforge.net/ , and includes  http://pandas.sourceforge.net/  , and  http://larry.sourceforge.net/ .

 

google spam filter 
  just a recurrent Warning Every once in a while google groups puts a message, even from an accepted member of the list, into the spam filter without notifying Skipper and me. If you send a message, and it doesn't show up on the google group within a reasonable time (and it's not spam), then ping me or Skipper... more »
By josef.p...@gmail.com  - May 17 - 1 new of 1 message    

pypi download statistic Robots ? 
  Just trying to see if anyone knows? I just realized that TravisCI is downloading packages from pypi at each run. does this count in the pypi download statistic? That could inflate the numbers pretty much, if other packages have a package as dependency. TravisCI is downloading zip files, which might explain some of the zip... more »
By josef.p...@gmail.com  - May 15 - 2 new of 2 messages    

TravisCI problems with python-dateutil - is there an apt-get? 
  We are seeing recently sometimes build failures on TravisCI because of pandas dependencies not being available. Is there a pre-packaged version for python-dateutil, that we don't have to rely on labix.org large web page to be available fast enough? ``` Searching for python-dateutil Reading [link]... more »
By josef.p...@gmail.com  - May 15 - 5 new of 5 messages    

Python/Stata 
  [link] "Reimagining a Stata/Python combination" I just saw this. A foreign language at the Stata conference ? Josef
By josef.p...@gmail.com  - May 15 - 3 new of 3 messages    

Python/statsmodels solutions for "An Introduction to GLMs", by Dobson & Barnett 
  I have now finished the Python/statsmodels implementation of all the code that is provided in R or Stata in the book Dobson AJ & Barnett AG: "An Introduction to Generalized Linear Models" 3rd ed CRC Press(2008) You can find the results on [link] As far as I can tell, there are three points where the statsmodels results... more »
By Thomas Haslwanter  - May 15 - 4 new of 4 messages    

Standard errors for Poisson regression 
  By now I have a guilty feeling, costing you guys so much time in going through my comments. Please tell me when to stop! Problem This time it is the standard errors for Poisson regression that don't match up. The fitted parameters, as well as the "fittedvalues" are pretty close (I guess that the mis-matches are just differences in when to stop the... more »
By Thomas Haslwanter  - May 15 - 8 new of 8 messages    

Applying function to a Pandas Dataframe with a MultiIndex 
  I would like to retrieve the rows of a DataFrame with a MultiIndex index, based on a given criteria. For, example, if I have a DataFrame that has observations for multiple labels on the same day, I would like to select the row or rows that meet a given criteria. I the data below I would like to select A12 until November 6th and then use B12 after that. One way to do... more »
By Aquil H. Abdullah  - May 14 - 3 new of 3 messages    

errors in for nominal logistic regression? 
  In comparing the output of the statsmodel MNLogit function with results from R and STATA, I ran into two questions/problems: 1) While the absolute values of the fitted parameters and the standard errors match, the sign differs from that in R: the R-command "multinom" from the library "nnet" res.cars=muninom(c_resp~factor (c_age)+factor(c_sex), weights-freq,... more »
By Thomas Haslwanter  - May 14 - 23 new of 23 messages    

Ordinal logistic regression / proportional odds ordinal regression 
  Is it possible to do an ordinal logistic regression (proportional odds ordinal regression) in statsmodels? I have not been able to find anything in the docu.
By Thomas Haslwanter  - May 13 - 6 new of 6 messages    

nominal logistic regression with formula/patsy 
  Is is possible to do a nominal logistic regression with a formula/patsy in statsmodels? And what about setting the frequencies? In R, one can set the frequency in "multinom" (in the library "nnet") with the parameter "weights=freq". In statsmodels, I currently somewhat clunkily hand-generate the endog- and... more »
By Thomas Haslwanter  - May 13 - 2 new of 2 messages    

1 - 10 of 1655   « Newer | Older »

XML       Send email to this group: pystatsmodels@googlegroups.com