This is FiveMinuteFriday, episode number 260, Data Science in Real Estate.
Welcome back to the SuperDataScience podcast ladies and gentlemen, super excited to have you back here on the show. Today we’re continuing our little saga or series of episodes on data science in various industries. So we’ve already covered off quite a few exciting and interesting industries such as banking, construction, transportation, education, entertainment, government, healthcare, mining and retail. And you can find all of those in the podcast archives at www.superdatascience.com/podcast. And now we’re moving onto the real estate industry. All right, so let’s have a look at this.
The real estate markets consists of a buying, selling, and renting property for commercial and personal household use. And it is a market that should not be underestimated because the annual revenue is in the trillions and the projected value for the annual revenue by 2025 is actually 4.2 trillion US dollars. As you know from our podcast, if you’ve been listening to these series of episodes, not many industries hit the trillion dollar mark and real estate is one of them. So it’s quite a large industry. And here are a couple of examples to put it into perspective.
So if we take UK, so the United Kingdom alone and look at only residential property transactions and we only look at the ones that are 40,000 pounds or more. So we don’t consider the small transaction. This is to give a perspective of how much data exists in this industry. So we’re looking at UK only residential property transactions and not commercial, and 40,000 pounds or above. There’s a report which we will link to in the show notes if you’d like to check it out, that has calculated how many of those transactions happen. It’s a staggering amount. 100.000 transactions happen per month. So hundred thousand transactions, residential property transactions or 40,000 pounds or more happen every month. That’s about 3000 transactions per day or 140 transactions per hour or over or approximately about one transaction every 30 seconds. So this podcast has been going on for about almost around two minutes now. So that means in this time 4 property transactions of 40,000 pounds or more have happened in the UK residential property market alone. Imagine what the number is globally and hence a huge amount of data, about property, characteristics, location, owners, buyers, sellers and so on. And that means there’s a lot of opportunity for data science.
Another interesting note to put into perspective is that the global real estate value is, so not the market transactions, the revenue of the market, but the actual value of real estate worldwide is over 200 trillion US dollars, which is over 250% of the global annual GDP. All right, so now that we have a good feel for the industry, let’s dive into the use cases of data science and artificial intelligence in real estate.
Use case number one, increased appraisal accuracy. So as you can imagine, getting accurate and complete information about real estate is extremely important to finalizing a deal, whether it’s for the buyer or a seller or somebody who’s renting out the property.
And that involves a lot of data. And in fact a lot of unstructured data, photos, videos, descriptions, written texts and so on. And all that needs to be somehow processed then. Companies are increasingly turning to data science to bring all of this data together and to bring it to their customers. And basically that is a way to increase transparency and therefore reduce risk for all the parties involved. The example here is Zestimate, which you guessed it is used to estimate the value of an individual home. The tool runs a series of processes built using various tools, mostly r but also python and others to analyze public and user submitted data like special features, location and market conditions. This allows them to estimate the sale and rental value of around a 100 million homes, 100 million homes across the US three times per week as well as to predict these values one year into the future. As you can imagine, this can be a killer application and can really help people out save a lot of money. And of course, it’s a very, has got a lot of potential.
Use case number two, smart homes. We already quite accustomed to the idea of Internet of Things and we wear smartwatches, use phones to control things and concepts like that. However, what is important to understand, and sometimes we might forget, is that the applications we wear on our bodies will only be a very small part of the whole Internet of Things with the majority of devices and sensors that will actually be around and in our homes, offices, entertainment and shopping areas. And that represents a lot of opportunity for data science to further enhance the value and functionality of real estate. An example here is Pointgrab, which is a workplace optimization platform that tracks behavior and space utilization of occupants in real time while keeping the tracking anonymous to ensure individual privacy.
And the result is that it claims to be able to reduce real estate expenditure by 30%. Staggering number, especially if you’re a business owner you can imagine how expensive real estate is to rent for your office. But however many of the features are still in development phase for Pointgrab. But nevertheless, that’s already a great example of another innovative idea of how to enhance value or reduce costs of real estate using data science.
All right, use case number three, recommendations. We spoke about recommendations earlier in our episodes on retail and also on entertainment, on those two industries. Well, real estate businesses also utilize data science in a similar way. It’s important to understand what a customer wants and what they’re looking for. And if you think about it, the average person only conducts several transactions per lifetime in terms of renting a place or buying a place or selling a place so the amount of data there in terms of the quantity of transactions is quite limited.
So we face that problem of small data that, Mike Segala was talking in one of our previous episodes. And so how do you approach that? Well, of course, the idea here is to create user profiles, which can be determined by other data that’s available, for instance, preferences and purchases in other areas such as clothing or travel, and then create these clusters or groups of users and then propose to them certain things about real estate. And we have an example here. Trulia Insight was launched in 2012 and has since gathered over two petabytes of data, with another terabyte coming in every single day. Integrating this tool in 2015 increased Zillow’s earnings by 7 cents per share, and their stocks grew by 55% in 2016. So if you’re not familiar with Zillow, Zillow group is a large real estate database company, which is actually listed on Nasdaq, and its revenue is 1.1 billion US dollars as in 2017.
So as you can imagine, it’s quite a big company and just integrating this one additional tool allowed them to grow their earnings by 7 cents per share and contributed to their growth in 2016. Data Science is super powerful bottom line. But yeah, as you can imagine, that’s data science solving the problem of small data in real estate.
All right. Example or use case number four, lead identification. Similar to recommendations, user data can be used by real estate agencies and banks to determine which customers to pursue. So the idea here is that banks are competing to provide loans to people who want to buy real estate. And the example that we have here is the company named simply First. So just the word First. Uses behavioral data from over 200 million people in the US to help real estate agents identify those looking to sell property in the near future. They guaranteed to predict about 15 listings per year for each agent, with about six to nine months before the seller ever seriously considers putting their property up for sale.
How crazy is that? It’s like one of those situations when you go on Facebook and you get an ad for sneakers and you’re like, yeah, that’s right. I need new sneakers. How did they know? Or even, so it tells you what you want or predicts what you are going to want even before you want it. Or when you go on Netflix and you get recommended a movie and then you end up really loving that movie or TV show even though you’ve never heard about it before. So kind of similar situation, probably even more advanced than that. And basically, what they’re using is, again, they’re using other information that’s available on people and by clustering methods, they can identify the best leads that will allow agents to directly contact the interested parties. So another pretty cool application of data science, predicting the future.
All right. And, our final use case number five, automated property management. So property management still largely functions in the same way as it has for decades. And that’s actually a great telltale sign. When you see an industry or some kind of service or product that hasn’t changed in decades, then it’s probably right for disruption. Something is going to happen there, there is opportunities there because things are changing so fast around. So well, property management hasn’t changed in a long time and now artificial intelligence is coming into the picture. And the whole idea here is to use AI assisted communication for managing properties. The example here is Zenplace is an online solution that combines numerous features for property owners and landlords. It uses a chat board via Alexa to answer attendance questions, reducing the landlords’ workload. They’re currently servicing properties with a combined value of over 500 million US dollars.
So as you can see from this example, it’s still quite a juvenile. It’s a very straight forward simple application. But that just means there’s a lot more room and opportunity for anybody to come up with ideas of how to automate property management or enhance productivity management even further with data science. But even this application is already providing value to people. As you can see, they’re servicing half a billion dollars’ worth of properties.
All right, so those are our five use cases of data science in real estate. Hope you enjoyed this quick rundown. If you are in real estate, then these are things to think about and see, where you can maybe apply your data science skills. And if you’re not in real estate, use these examples as a way to spike some new creative ideas in the industry that you are in. Because data science is a very transferable skill sets and tool. And so if something’s happening in one industry, there’s nothing to say that it cannot be used or a similar idea, cannot be used in a different industry.
On that note, thank you so much for being here today. You can find all the show notes at www.superdatascience.com/260, that’s a www.superdatascience.com/260. We’ll include all of the links to the materials that were used for this research, as well as the transcript. And with that, we’re going to wrap up. Once again, thank you for being here today. I’ll see you next time. And until then, happy analyzing.