I have users on our online platform which are not evenly distributed across. In identified users, we have 5M Female vs 3.7M Male. Which means if I want to indetify new user with out any previous history on website I will be always biasing towards female
My question is how I can
- Go back to the old look-alike approach and check if and how that was debiased.
- Also sorry if I can not formulate my question well
Any suggestions/help will be appreciated
Thanks in advance