They're (bigger pockets) really out of touch with what is happening with real estate data, or how it can be used. BiggerPockets is all about the 'clicks' and SEO from the watered-down articles. When 'ever' have you seen one of the cheesy webinars touch on anything relating to Jupyter Notebooks (Open Source = free) or true analysis of the sale/property data? It's been the backbone tool of the hybrid RE disruptors coming out of Silicon Valley stomping the bejeezus out of the average real estate agent
There IS a disconnect between the real estate professionals and data science folks. The data folks know how to 'load' the data (it's actually really easy to do), but use the same generalized data sets in many of the housing models that get national press for sales and predictions. They use the generalized data sets.... because many of them have no clue where or how to get the property data that matters.... and on top of that, they do not realize the significance of the data points they are looking at, and remove them from the models.
On the flip side of that, most folks in organized RE, (licensed agents) are actually pretty tech illiterate (due to the oh-so-very-low educational bar set to become licensed), if it's not a tool or data source spoonfed to them from a local MLS. But they do understand the variables (the last sale deed type, and $ amount of the last sale, local housing plans/initiatives ect) that actually DO contribute to being able to predict housing prices.
Here's a recent prime example of a solid approach to comps
https://towardsdatascience.com/predicting-house-prices-with-linear-regression-4fc427cb1002
She goes on to say 'this data set contained over 20 categorical features' which is kind of a joke. See anything wrong with that? But here is your blueprint to work off of.
Locally where I am in Mpls, I can find a very 'clean' 250 columns of data to load into a Jupyter Notebook consisting of if the parcel is a licensed rental, property type, what recent housing/tax plan that it is part of, deed/sales information, 'attributes' of the structure (beds/bath/sf ect) and it's 'free' data... and ya know what? You can similar in every major metro, if you 'look' for the data.
Do a search on BP for 'Jupyter Notebook'....
---
I think that's way out of scope for this site, at least the way its presented in that article. That's asking Real Estate agents to tune Machine Learning models. But to your point, higher level discussions and even just information of what's going on with Real Estate Data would be a big plus.
That said, I've got a lot of questions about this because I'm a programmer. :-) Going to shoot you a DM.
Thanks for sharing.
