Your Data Science Cheat Sheet
People often throw around terms like Artificial Intelligence (AI), Machine Learning (ML), and Predictive Analytics interchangeably in conversations about new real estate tech.
AI-powered “bots” that serve as an agent’s personal assistant.
Products that “learn” how to respond to leads.
Predictive models to identify who’s going to sell and when.
These technologies hold enormous possibilities, which is why I think it’s important we all understand what they can do and what they mean for the business of real estate. Agents and brokers will become more efficient, and more productive, because of them.
But despite all the buzz, there’s a lot of confusion about what these things actually are, how they relate to one another, and how they can be put to work. You may have wanted to ask, “well what actually is that?” or “are those really the same thing?” Even a quick Google search may leave you more confused than when you started.
I therefore present you with a simple “cheat sheet” that gives you the basics in clear, nontechnical language:
Artificial Intelligence (AI):
The broad umbrella term for intelligent systems driven by algorithms — software that can perceive inputs, learn from interactions, and optimize to specific outcomes.
What it IS: the frontier of machines doing things we once thought only humans could. Best poker player in the world? It’s a computer. Diagnosing lung cancer? Computers are more accurate. Recognizing faces? Yep, Facebook’s algorithms beat human performance now too. Driving? We’ll see soon!
What it ISN’T: terminator. General purpose intelligence that will replace you (or even your ISA) won’t be here anytime soon. The current state of Chatbots is a good example of the limitations of these layers of “intelligence”. Many AI companies have humans running in the background until they build up a large enough data-set to accurately handle more use-cases. It’s easy for AI to look sloppy.
Machine Learning (ML):
One foundational branch of AI — the core algorithms and statistical methodologies that enable machines to learn on their own, not taught by humans. Machine Learning is essentially statistics on steroids.
What it IS: freaking complicated. Seriously, the junior Data Scientists on our team have PhD’s in experimental particle physics and can code to boot. Decades of statistical and computer-science advancements + massive compute power + reams of data = enormous business opportunities for products that could not have previously existed. It’s why “Data Scientist” was the #1 hottest job in the US in 2016.
What it ISN’T: magic. Machine Learning requires 1) massive data, and 2) incredibly talented oversight. Designing data science experiments that train machines to iteratively improve performance on key business problems requires a huge amount of human input and effort. However, the results can be pretty darn magical.
Any solution that uses past data to predict (estimate) a future phenomenon. Can be as simple as “if-then” rules, regression analysis, or leverage the most complex Machine Learning techniques.
What it IS: A primary objective of business analytics across the globe. Predictions empower businesses to optimize marketing efforts, forecast success rates of new products, set prices and more. You know how Netflix suggests movies you might like, Amazon knows what you might want to purchase next, and Zillow prices homes (and is paying $1.2M to improve their model)? Whatever the current accuracy, those are some examples of Predictive Analytics.
Predictive analytics can also be remarkably simple — if you ever built a forecast with the goal of projecting future sales, you were doing predictive analytics.
What it ISN’T: new. In some cases, new access to broad data and powerful tools enable significant increases in predictive power. The challenge is that people can slap “Predictive Analytics” on their website for just about anything — doesn’t necessarily mean there’s sophisticated software or meaningful value.
What the hype means for real estate
Full disclosure: I run a venture-backed RE company with a product powered by machine-learning driven AI, so I care deeply about helping agents cut through the hype of data science jargon. My hope is we can get beyond the techniques — AI, ML, etc — and focus on results.
I believe the most exciting opportunities for data science in real estate are to supercharge what agents are already doing — without ripping out the heart and soul of what makes real estate great. Forget Robocop, think Ironman.
Do you know the average time pilots actually FLY commercial planes thanks to what was once the forefront of AI? 7 Minutes. But I prefer my flights to have pilots (otherwise it’s the train for me, thank you very much).
Has the pilot’s job gotten a lot easier? Heck yes. Are they still essential? Absolutely.
I’m excited to see data scientists and engineers flooding into real estate to help you pilot your business into the future.