Statsmodels gave me an interesting comment
Possibly complete quasi-separation: A fraction 0.27 of observations can be
perfectly predicted. This might indicate that there is complete
quasi-separation. In this case some parameters will not be identified.
Has this happened to someone before? It's kind of new thing that has happened to me
Hey Avtandil,
Thank you for your question!
Such a message occurs whenever the target is completely or partially determined by some feature in the dataset. You are welcome to research the various techniques to deal with the problem :)
Kind regards,
365 Hristina
It is working without this fix now
#Apply a fix to the statsmodels library
from scipy import stats
stats.chisqprob = lambda chisq, df: stats.chi2.sf(chisq, df)
and
got the same comment when I gave
results_log.summary()
