jamesdj / tobit Goto Github PK

Hi James Jensen,

I have a question regarding the Tobit regression file. The program estimate the regression coefficient of the Tobit regression, however it does not say anything about the significance of the regression coefficient.

My question is: Can the significance of the regression coefficient also be estimated using this package?

Best,
Tom

Intercept extraction index in wrong

On line 156 in tobit.py.
The intercept is improperly extracted from the result of the optimisation, it is saved as:
result.x[1]
which the first of the coefficients, not the intercept.
It seems it should be:
result.x[0]

The index should be 0, not 1.

if self.fit_intercept:
            self.intercept_ = result.x[1]  <- THIS LINE HERE
            self.coef_ = result.x[1:-1]
        else:
            self.coef_ = result.x[:-1]
            self.intercept_ = 0

Hope this helps.

Hi James
This code is really helpful. It runs well for 1 million rows. However it is not running if I load data for 30 million rows.

1. Could you please advise if multiprocessing is possible for this code? will you be able to incorporate that in your code?
   OR
2. Could you please provide a similar code for Tobit which runs in pyspark

Regards,
Shaju