AI & Real Estate: Part 2
Learn what we’re doing to revolutionize the AI/PropTech space, and find out what they aren’t doing.
This is part 2 of a 3 part series. Part 1 can be read here. Part 3 is coming soon. Follow us if you want to read more.
—
Our Data Journey
In Part 1, we touched on a few introductory AI concepts and some key foundational principles of AI processes. One of the most essential principles covered is that AI and predictive analytics can only be as good as the data your working with. Simply put, data that is outdated, fragmented, skewed, or unorganized can hinder and impede accuracy and effectiveness of the algorithms and systems based on this data.
At Estated, we’ve dedicated a massive amount of our time and resources to continue sharpening our own AI processes in our back-end models and inside of our products. This of course starts with the process of collecting and organizing data from several sources, and then making them all work together in the same algorithms and programs.
With AI and Machine Learning frameworks becoming more and more commonplace in PropTech and it’s affiliated sectors, this has continued to become a priority for us as we grow and plan ahead.
Currently, AI is most effectively utilized in our AVM technology and our Comparable Report.
What is an AVM?
With property valuation being an integral fixture in the housing industry, it has been a fixture long overdue for disruption. AVMs, or Automated Valuation Models, are intelligent programs that automatically analyze sets of data to produce an estimate on the current value of a home or property.
Most often, online visitors type in a property address and the engine uses linear and multiple regressions to form an estimate of that property’s market value. These datasets can include the property value, market values, trends, historical data, and the age of a home.
The more variables at play, the more specific and precise AVM estimations can get. The graphic above visualizes the composition of the layers and their interaction with each other. It also shows the grouping of similar variables to increase their potency and intelligence, making the layers much more effective than more simplistic predictive models rooted solely in sales trends.
What makes a good AVM?
The current state of AVMs in PropTech is commonplace and competitive. Current AVMs can be found among most property report services, but they tend to fall short of the mark in many ways. Zestimates, Zillow’s infamously inaccurate AVM system, is perhaps one of the best examples of this.
The Zestimate remains an inaccurate valuation model Standard real estate pricing models always take into account recent sales into the pricing a home, and Zillow is no different. In fact, the amount the sold price is pretty much the end factor of deciding what a home is worth. But even at that, the Zestimate can be skewed all kinds of ways. For example…
Pictured above is the home of Zillow CEO Spencer Rascoff who sold his house for $1.05m while his Zestimate had it at $1.75m, 40% over the selling price. Our valuation, pictured below, had it much closer at 1.3M.
The bottom line is, the Estated valuation is rooted in more comprehensive data, it implements cutting edge machine learning techniques, and models the value of a property much more naturally.
Comparison Reports
Our new report feature, the Comparison Report, compares the property in question to similar homes nearby using AI processes to match property features, proximity, price, and other aspects.
While this isn’t exactly groundbreaking, the development of this report will allow users to interact with and find properties based on filtering by property characteristics altogether, similar to e-Commerce filters.
In the report, we also break down the valuation in a year forecast, estimate mortgage prices and estimate rental income, all based on the aforementioned AI processes that comb through similar properties.
What’s Next?
In Part 3, we will inspect AI and real estate further by looking at it’s future potentialities. Stay tuned to find out exactly how AI will revolutionize the home-buying process in the years ahead.
This is Part 2 of 3 pieces exploring AI and it’s current place in real estate technology. Follow us if you want to read more.
Questions? Comments? Great GIFs?! Find us on Twitter here.
