Big Data has long arrived in the real estate industry. Digitization and the intelligent use of large amounts of data will fundamentally change the industry, expect experts such as Juri Ostaschov, chief data scientist at Prea Group, a real estate investment advisory and service company. A conversation about clean data, simple structures and forecasts.
Mr. Ostaschov, what role does big data play in real estate investments?
Juri Ostaschov: Anyone can use big data to obtain relevant information at the smallest geographic scales. Some can only be analyzed at the postcode level, while other data can be analyzed at the object level. Real estate developers have quick access to framework conditions, for example what they can do exactly where according to the development plan. Investors get a good overview of the costs, where there is how much rent per square meter, why the prices are at a certain level and whether the rent invested can still be achieved sustainably for the next five years. Asset managers can learn about the risk of potential default at an early stage. Such results based on big data surveys are used to optimize the investment performance of clients and, for example, to better structure equity and debt capital. A current acute question could also be: how much is a neighborhood affected by Corona?
Open detailed view
Juri Ostaschov, Chief Data Scientist of the PREA group
What data do you access to provide such answers?
We use many different sources. This includes data that is public anyway. Also of interest are online databases such as GeoMap. We also buy data from research institutes, among others. Once we have found the sources of certain questions, the data needs to be cleaned up, transferred, and put into the proper context. It requires computing power. We can query up to two terabytes in 20 seconds. But that’s not enough, you have to do something qualitative with the large amounts of data.
What does qualitative mean?
The art of data analysis is to filter the desired responses from gigantic stocks. We work with many sources that seem to have nothing to do with each other: for example, the density of restaurants and the way they are rated on certain portals can provide information on the development of a trendy neighborhood. This is then coupled with the question of how the economy develops in general. Converting all of this into reliable data is the challenge. As the first purely digital investment advisor in the real estate sector, Prea has developed its own AI for this purpose. It is not a question of using the most complex algorithm possible, simple structures make a system more secure. In order to always find the best path here, I develop the AI, but I am also responsible for the analyzes.
What can big data do that was not possible before?
Data sets based on a large geographic basis could not previously be focused on local points. It is now possible thanks to AI with big data. It becomes interesting, among other things, in places with little traffic. Today, it is no longer enough for us to look at five comparable properties and make a forecast based on them. Our base is larger because we bring together such “comparables”, no place in Germany is unique. Clustering algorithms here help analyze and filter locations for similarity. It is crucial to be able to access the right sources to the extent required. This data can be used to create exact simulations. This is the second big step forward: Big data enables precisely quantified statements, much more precisely than vague predictions of trends.
What information does this allow beyond abstract numbers?
We performed big data analysis for the seven largest German cities: You learn things that may not have been considered before. In Berlin, for example, parquet floors and an open kitchen are very advantageous. In Düsseldorf, on the other hand, it is more important that the apartment is non-smoking and that it has a full cellar. In Stuttgart, on the other hand, very favorable price factors are a parquet floor and a roof terrace.