statsmodels OLS fit computation time

Question

I am using statsmodels.api for some simple OLS regression... And somehow every time I ran my script it got stuck at model.fit and I couldn't figure out why.

Here is a snippet of my code:

import statsmodels.api as sm

merged is a pandas data frame as regressors and memoscore is a pandas data frame of one variable as my dependent variable. The following worked smoothly and instantly:

model = sm.OLS(np.array(memoscore), np.array(sm.add_constant(merged)))
results = model.fit()

Then I took the log of memoscore , the following still worked instantly:

memoscore_ln = np.log(memoscore)
model = sm.OLS(np.array(memoscore_ln), np.array(sm.add_constant(merged))

But it got stuck here forever:

results = model.fit()

Could anyone kindly suggest reason why and/or how to get around that? Thank you so much in advance!

Answer 1

I exported my data to R and ran the same OLS using memoscore and logged memoscore as the dependent variable, it worked like a charm. Still don't know what's wrong with statsmodels , but at least I learnt R is the go-to software for such easy regression tasks.