0

So I am running a statsmodel code for Logit to compare statistical significance. I am doing one on one classification. My model results return duplicate answers for the same variable with at times opposite coefficients( negative for one and postive for the other) and opposite signifcance p vales ( higher than 0.5 for one and less than 0.5 for the other.

Here is an exmaple:

                                coef    std err z   P>|z|   [0.025  0.975]

top_notes_me[T.orange (bitter)] -0.2234 0.253 -0.882 0.378 -0.720 0.273 top_notes_me[T. orange (bitter)] 0.7070 0.199 3.555 0.000 0.317 1.097

The p values here are different and so are the coefficents. Can anyone help me understand what the problem is? My guess is there could be spaces in the final dataset with slight difference that are treated like this here.

I haven't tried anything as I am not able to understand what exactly is wrong.

user20479548
  • 1
  • check the dataframe of your data. My guess is you have some extra spaces in your oranges. – Josef Nov 11 '22 at 17:38
  • Welcome to SO! I'm not sure, but this might not be an [MRE](https://stackoverflow.com/help/minimal-reproducible-example). You may need to [edit] more info into your post. – rainbow.gekota Nov 11 '22 at 21:16

0 Answers0