You have a dataset with 11,000 observations and 21 predictors. You know from Domain knowledge that some of your predictors have interactions with each other.

What would be the best way to check for interaction terms within your model?

a)Add all possible two way interactions and then use Stepwise Regression to eliminate non-significant interactions.

b)Use your Domain knowledge to add and check interactions between a few predictors.

c)Create a correlation matrix and add interaction terms for all predictors that show a positive correlation.

d)Create a correlation matrix and add interaction terms for all predictors that show a negative correlation.

e)None of the above.