[loftyai]

Yeah data is always a big headache haha, so I feel your pain.

We use MLS data to obtain housing prices historically. We don't use government data for median income, instead, we estimate it directly by tracking social media posts from the neighborhood that are public.

We then run image analysis on them to detect features like the types of dogs people have, types of cars, and other stuff. It's not 100% accurate for sure, but it's given us a pretty good understanding of the median income for neighborhoods, and the data refreshes in real time too :)

[rog211]

Very cool thanks for the insights! To run you model you must have some baseline data to be able to train against median income. How do you do that?