Case Study – Union Investment and Geospin develop automated location evaluation

Challenges of the real estate industry
One problem of the real estate industry is its lack of transparency, which is reflected in a delayed and incomplete data basis. For example, data on prime rents and ground values are often incomplete and time-delayed (Schüppler 2019). The collection and interpretation of data takes a long time in the conventional way and often delivers imprecise price determinations due to the lack of data. Many real estate companies therefore still make decisions based on a combination of intuition and retrospective data (McKinsey & Company 2018).
Automated location evaluation – implementation of a proof of concept
Union Investment decided early on to take advantage of new technologies and commissioned us to conduct automated location evaluation in June 2017. As part of the proof of concept, various methods of machine learning were used to test their applicability for the automatic location evaluation of office buildings. The square meter rental price served as the target variable for the individual models.

„The approach highlighted how automated location evaluations can support our internal processes. A particular challenge comes from the availability of large and constantly updated datasets.“
Dr. Peter Scibbe, Senior Business Expert, Union Investment Real Estate GmbH

Using Hamburg as an example, we showed Union Investment that big data and machine learning methods make it possible to automatically evaluate office locations using freely accessible geodata.
A major challenge for meaningful results is the availability of extensive input data. In this case, input data was provided by ImmobilienScout24 to ensure an adequate database.

Added value of our automated location evaluation
Our location evaluations take into account the complex surrounding structures that influence the rental and purchase prices. This allows us to identify correlations between property prices, existing infrastructure as well as hard and soft environmental factors. It is hardly possible to identify such relationships using conventional methods. Our algorithm is scalable and can also be applied in regions for which only few comparative values or market information is available. A further added value of the solution lies in the fact that future rental price developments can be anticipated.
Immobilien 3
This heatmap shows the area-wide prediction of the square meter rental prices of commercial properties in Hamburg. In addition, the reference values of ImmobilienScout24 are shown. The rental prices of commercial properties in Hamburg predicted by us only deviated by cent amounts from the actual market prices.
Our procedure simplifies work processes and contributes to a better assessment of the potential of investment properties and increased transparency.

– McKinsey (2018): Getting ahead of the market: How big data is transforming real estate,
URL: https://www.mckinsey.com/industries/capital-projects-and-infrastructure/our-insights/getting-ahead-of-the-market-how-big-data-is-transforming-real-estate [02.01.2020].
– Schüppler, Ulrich (2019): Für Asset Manager wird eine Vielzahl von Daten wichtig, in: Immobilien Zeitung,
URL: https://www.immobilien-zeitung.de/150076/fuer-asset-manager-wird-vielfalt-von-daten-wichtig [02.01.2020].
– cover picture URL: https://www.schluesseldienstvergleich.eu/


Contact  & follow us