I am working on a project and I wanted to get some thoughts from the group about using gender and age as predictive variables for a model. A few times that I could see definitely no problem is in the medical field, but what about when working on a customer churn model or something similar? Any rules of thumbs that people use when deciding which data to include and which to leave out?
Not really a solution but maybe interesting:
From a model perspective every feature that improves is helpful.
From an ethical and cultural perspective this is much more trickier as you have to define them in the first place and they can change depending on the place or time.
In your case for age and gender I would assume they are commonly used features today?
They are at least commonly known features. We aren’t doing a ton of analytics overall at the moment.
@KevinB, I think you have to start by thinking in how this model will be used and how a sex/age bias can/cannot affect lifes in this usage. It is hard to give more thought without understand the context. There are no moral constants.