If I perform a linear regression in R, I get a nice summary of the resulting model, $R^2$, p-values for different features, etc.
If I do the same in scikit_learn, I get nothing of this. Are there any ways to print summary of the model there?
Asked
Active
Viewed 6,358 times
2
-
possible duplicate of http://stackoverflow.com/questions/26319259/sci-kit-and-regression-summary/26326883#26326883 – eickenberg Oct 15 '15 at 14:27
-
2as a general comment: scikit-learn is for machine learning, not for performing classical statistics. It is predictive power of the model that counts, not statistical significance. Totally useless predictive models can be statistically significant. – eickenberg Oct 15 '15 at 14:29
1 Answers
6
Scikit-learn does not, to my knowledge, have a summary function like R. However, statmodels, another Python package, does. Plus, it's implementation is much more similar to R.
from statsmodels.formula.api import ols
#you need a Pandas dataframe df with columns labeled Y, X, & X2
est = ols(formula = 'Y ~ X + X2', data = df).fit()
est.summary()
ansonw
- 1,559
- 1
- 16
- 22
-
2Yes, I am aware of existence of statsmodel. But may be there're some external packages which might provide similar interface to scikit learn? – Konstantin Solomatov Oct 14 '15 at 23:06