A Data Science Central Community
What is the significance (or consequences) of not including interaction terms of variables in logistic regression?
I have 27 explanatory variables. How should i decide for which pairs of variables, I should introduce interaction terms.
Can i rely on my domain knowledge e.g. income and employment type are correlated when predicting default prob. and hence i should introduce interaction terms for these two?
Any inputs on this would be helpful as I am a newbie to analytics.