I have a dataset with 10 columns and 158 rows. I try to predict my test dataset which is 1 columns with 158 rows.

I made cross-validations, grid-search and use ElasticNet algorithm.

Also before the evaluate the model I check the pearson correlation between 10 columns which I used for train with other 1 column which I try to predict. The correlation is not good but when I evaluate model the R^2 score is near 0.98 .

How can I be ensure that this score is confidental ? Because I didn't expect a R^2 like this. This is too high that I expect.

Thanks in advance.


$ R^2 $ shows what variation of your purpose variable is described by independent variables. So their synergetic effect could give you good better answer than their correlation. Better use $ R^2 adjusted $. Look at p-values of your variables and think about their real correlation. Are they important in real life. If they have any adequate relationship, so your regression is right.

Hope it will help.


Try regression diagnostics. There are a couple of methods to help you make sense of the data and the model. Try this link, go to resources tab and download Chapter 6 - Diagnosing Problems in Linear and Generalized Linear Models. The code is in R. However you can find the python equivalent of the code.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.