The data available through Open Data is a gold mine for professionals who know how to handle it.
The digitization of the economy has generated huge amounts of information about people, their behaviour, goods, exchanges, values… This is called Big Data. At the same time, the Open Data policy has made many databases previously reserved for administration accessible to everyone.
Data mining is the set of techniques that enable the exploitation of these databases and Big Data. Data mining is an Anglo-Saxon term that can be translated as “data exploration” or “extraction of knowledge from data”. Specifically, it is a family of automatic or semi-automatic tools that allow the analysis of a large amount of data. The goal is to find correlations, anticipate trends and make decisions based on reliable information. Data mining is primarily a decision-making tool, explains William Violet of Homiwoo. This is confirmed by Ludovic Gauvin, director of data at Yanport, whose job is to structure data and build relevant tools to help real estate professionals make decisions.
Real estate is particularly concerned
The existing data useful to real estate professionals relates to properties, prices, advertisements, people, territories… For people, we can know their gender, their age, their socio-professional category, but also their behavior, especially on the Internet and on social networks. All these databases fit well with the definitions of Big Data, they are very large and Data Mining is necessary because their exploitation requires complex tools. The number of interesting real estate bases is growing: “Over the past three years, we have had more and more visibility into transactions,” says Pierrick Pretot, CEO of Le Prospecteur.
Know the values, look for mandates, conduct market research
The concrete applications for real estate can be grouped according to three generic functions: knowing the values, looking for mandates, conducting market studies.
Knowing the values, sales prices or rents makes it possible to make the assessments more reliable, but also to offer online estimates to identify the owners who have a real estate project. The search for mandates, in addition to online estimates to generate leads, is done through behavior analysis on the Internet or behavioral databases. Of course, conducting market research allows you to know your area in order to better serve your customers, as well as to publish rich, relevant and current content on a website, a blog or on social networks.
The databases used are kept secret by the companies that manage them, they are part of their business. But they all analyze the DVF file, Insee, Data.gouv.fr and the announcements. Some have agreements with banks and insurers to use their files. For William Violet of Homiwoo, there are three main use cases: capturing leads, professionalizing value advice, and running market observatories to do SEO. For Ludovic Gauvin from Yanport, in the future we will also be able to use the technical data of the building, for example BIM, to refine the evaluations and we will have advanced indicators by analyzing data from Google Trend.
The methods of analysis are very complex and the teams are made up of high-ranking engineers and mathematicians. One of the companies has partnered with the École polytechnique to develop its analytical methods. The complexity stems from the fact that the data is only interesting if it is crossed with other data. Pierrick Pretot describes the Data Mining process as follows: “The data is recovered by robots, APIs or in Open Data, it is then stored on special servers, then processed to be readable and finally crossed with other data to speak”. The Prospector retrieves over a hundred events or signals that are precursors or triggers of transactions.
Start-ups that grow very fast
The companies that offer data mining to real estate professionals are young start-ups, with one exception. They are rapidly evolving and changing their service offerings for greater precision and to better meet the needs of brokers.
Starting up data mining
Don’t miss too