Transforming French Real Estate with Data Science
According to just about everyone data science is ‘the next big thing’ or, rather, is the big thing right now. In fact data scientists are so in demand right now that there are a far greater number of positions available than scientists to fill them. In fact, some are putting the number of data scientists needed over the next decade to be over one million.
Data scientists are the human component to the global phenomenon known as Big Data. Which is a nice descriptive word for the fact that many companies, governments, and researchers are generating data sets that are so big that entirely new strategies have to be developed for handling them. The field of handling, cleaning, and processing these giant data sets and then finding the value within them is called ‘Data Science.’
What exactly is data science?
To get a little insight into this relatively new and very broad field I talked to Guy Lifshitz, CTO and Head of Data at Homeloop here at Le Studio. “Data science is a really broad field” he says “a lot of things can be called ‘data science.’ It could be anything from the actual gathering of data itself to modeling based off of a cleaned up data set”
“And everything in between” he adds, “from cleaning it up to preparing data for use.” Data Science has found a home in health care, public transportation, athletics, concerts sales, basically in every industry that collects data. It has been especially eagerly adopted in marketing, advertising, and in the many sectors of the finance industry. Any industry or company that generates a lot of data over the course of doing business can benefit from implementing ‘Big Data’ strategies.
What can we use it for?
Having studied Computer Science at McGill University, Guy was quickly exposed to data science and the myriad of ways it could be applied to almost any field. “My first real exposure to data science was actually while working with the Political Science department,” he says, “and one of the first projects I worked on focused on examining political contributions in the United States.”
“The government records basic information concerning the origin of donations, but it doesn’t (or didn’t at the time) have a way of connecting donations that originated from the same person but came from different physical addresses. So, one could easily donate money from different address if they had multiple homes.”
This project was designed and constructed in the long shadow of the Citizens United verdict, so the public gaze fixated on political donations and the role that they were playing. “We found that quite a few people were over the technical limit!” he says.
Data science is applicable to pretty much every field, business, or science that, well, gathers a sufficient amount of data. “Another interesting example is predicting political opinion on social media even when the user doesn’t actually post anything political, he says, “using network effects one can extremely accurately predict the political leanings of social media users.”
Data Science in Practice
Homeloop is one of the major startups that we currently have at Le Studio. With the goal of being able to quickly and accurately price and sell properties in Paris and the greater Île-de-France region, Homeloop relies heavily on data science. In fact data science is the very core of the service.
“Right now we have an algorithm but we are trying to improve it by adding fields, or parameters that might affect value such as neighborhood crime rates, proximity to parks, and daytime traffic on the roads near any given apartment.” Guy says. A lot of the data which Homeloop ‘feeds’ its algorithm comes from the public domain, a kind of data referred to as ‘open data.’ However, “we have also bought a considerable amount of data to supplement the public data and help improve accuracy” he adds.
“We want to add fields incrementally as the algorithm already works as is. We’re trying to incorporate as many different possible parameters as possible. For example we are even going so far as to include modeling of access to sunlight based on the position of an apartment and its location in relation to the surrounding buildings.” he says.
To create and train such an advanced algorithm, Guy and the Homeloop team are utilizing a kind of machine learning. That is, by feeding their algorithm ‘training sets’ it can begin to tune itself in to the different parameters. It is then repeatedly and systematically tested and compared with real world results. The algorithm will then learn with experience and begin to accurately predict with greater and greater accuracy the ‘real’ price of a given piece of real estate just based on the available data.
“As the algorithm becomes more and more fleshed out, it will be able to automatically adjust prices for changes in neighborhoods. For example if a neighborhood is gentrifying or if crime rates are decreasing in a certain neighborhood, the algorithm will be able to automatically start adjusting prices up or down.” Guy adds.
This is just one of the ways that Le Studio is looking to take part in the data revolution. As more and more industries begin to digitalize and more and more data sets are made available to the public, we are going to be be looking to take advantage of these opportunities.
The internet revolutionized and continues to revolutionize the world. It and the exponential rise of storage and computing power has allowed us to amass enormous amounts of data. The new frontier is sorting that data and separating the wheat from the chaff.
Le Studio VC is a Paris-based startup studio. Providing seed capital, expertise, and guidance, Le Studio acts as a launchpad for startups. By both investing and participating in the initial stages of development, success is more likely and more rapid.
If you are interested in starting your own company or working with us — let us know at Le Studio!