Use the data to set the price of real estate

Bouygues Immobilier, a leader in private property development in France and in Europe, has been developing residential, tertiary and commercial real estate projects for its clients for almost 60 years.

To study the different projects and orient its marketing strategies, the company relies on its Datalab. A team dedicated to the study of real estate data that Inka, agapian since September 2019, joined as a Data Scientist Junior

Identify the Levers of Goodwill

Within Bouygues Immobilier, the data lab responds to requests related to the data, and in particular, carries out studies of real estate projects (from simple construction of housing to the creation of complete neighbourhoods). These ad hoc studies allow Marketing to determine a marketing price and guide their strategies. The goal is to define and analyse the levers of goodwill specific to each construction project. In other words, an intrinsic (characteristics of the project) or extrinsic (characteristics of the location) element that would positively influence prices. The data lab thus responds to specific requests internal to the company, which facilitates both exchanges and understanding of the issues.

Estimate the Marketing Price

I, therefore, have to exchange with several teams with varied characteristics to offer them lines of study consistent with the available data and the project. Once these axes are defined, I perform statistical analyses based on real estate data, and I seek to highlight results that are robust and consistent with reality. I, therefore, seek to quantify
each of the levers of goodwill and then apply them to property prices so as to estimate, for a given project and date, the marketing price. It is also essential for me to seek data useful for the study of these projects. It can be geographic data (where are the possible assets?), Economic data (how much are these or that housing, in such or such place) or scientific articles on real estate studies. At the end of the study, I present the results and the estimates thanks to visual support which must be both simple and understandable for trades which are not necessarily familiar with statistics, data analysis or even ’computer science.

Each study is different

From a technical point of view, the quantity of data analysed requires the use of appropriate tools, in particular Pyspark, which allows parallelisation of calculations and thus avoids time-consuming processes. I also work on a collaborative analytics platform, Azure Databricks, which offers significant analytical power and the possibility of sharing my scripts simply with my colleagues.

Since the beginning of my mission within the data lab, I have been involved in a dozen real estate projects. What I like about this project are the challenges brought by the specificity of each study. Indeed, it is difficult to automate analysis processes in the face of the complexity specific to each request. It is, therefore, necessary to establish roadmaps and to exchange regularly with stakeholders. These exchanges are very interesting because they allow me to gain knowledge specific to the field of real estate and marketing, which, basically, are not part of my specialities.