I'm running a number of regressions I want to superficially compare with Stata, to ensure my code works as I migrate from Stata to Python.
i.e statsmodels.formula.api('x ~ y') outputs
coef std err t P>|t| [0.025 0.975]
------------------------------------------------------------------------------
Intercept 2.9420 0.287 10.265 0.000 2.380 3.504
Whereas Stata provides
------------------------------------------------------------------------------
med | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
_cons | 2.94203 .2866093 10.26 0.000 2.380268 3.503792
I'd essentially like the output to match exactly. I know this can probably done under the hood by saving the coefficients, but it's far easier at a glance.
The (hacky) way I have achieved this is by editing the statsmodels library file iolib\summary.py, and including Stata-like rounding. It's not perfect, but it will do the job for the moment.
def forg(x, prec=3):
if prec == 3:
#for 3 decimals
if (abs(x) >= 1e4) or (abs(x) < 1e-4):
return '%9.3g' % x
else:
return '%9.3f' % x
elif prec == 4:
if (abs(x) >= 1e4) or (abs(x) < 1e-4):
return '%10.4g' % x
else:
return '%10.4f' % x
elif prec == 10: #stata like coefficient rounding
return round(x,2)
elif prec == 11: #stata like t-stat rounding
return round(x,7)