Their indicators can be wildly off because they do not know the person themselves and they can only make guesses.
Also,
> Seriously?
Yes, seriously. Of course seriously. Stop being insulting and rude.
Not that I know what they do of course, but without much thought I can think of a model that would allow me to predict income.
- They're using Uber, that cuts out a decent chunk of poor people.
- What areas do you visit and at what times?
- Property prices in those areas, are they workplaces or urban commuter areas? Can likely predict value of someone's house if they use it from home.
- Type of Uber taken
- Size of tip
- Frequency of travel
- etc. etc.
I reckon that given some time and some people's ride data I could estimate their income, house value, and even perhaps job. I would guess Uber have done this and then some.
Of course it's not going to be 100%, but Uber don't need that.
Reading your response I suspect you failed to account for the fact that what "Uber knows" is not limited to the set of facts that you choose to disclose to Uber. They can and likely do use those facts to lookup your information in third party databases. They also know facts that you may not have chosen to disclose but unwittingly did so.
There is nothing to suggest Uber "[does] not know if you are poor", they could very easily be buying this data on their users.