Correct me if I am wrong, but from what I've been reading for Machine Learning models, it is the procedure to find independent input features that are correlated to your target variable.

However, what do we do when the target variable and the input variable are conditionally dependent?

For example: The target variable is having a Checking account. In order to open a Checking account, you must have a Saving account open. However, not all people who have a Saving account have a checking account.

How should this be handled, and should HaveSaving be included in the input features?

Edit: Revisiting this problem because I read somewhere that ML deals fairly well with the idea of multi-collinearity.

Edit 2: No responses yet ... am I missing something from the post people need? Would like to know that so I can restructure the question if needed.

However, I just can't shake the feeling that the model I am building is being influenced.

Am I assembling the data with the right mindset? I am afraid I am making the ML algorithm biased by supplying so many columns that are derived from Savings. I did feature transformation that got me several more columns relating to Savings.... will this influence the ML algorithm too much from a data perspective?

enter image description here


Your Answer

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

Browse other questions tagged or ask your own question.