The challenge
The customer is a real estate company operating in the German market. It wanted to monitor and predict price dynamics across residential and commercial property, to value properties accurately and automatically, and to discover attractive investment properties automatically.
Our solution
We built a data collection pipeline to analyse over one million real estate items and trained a deep learning model on that data. An interactive visualisation tool gives users straightforward access to the model’s insights.
Although the tool was built for the German market, the approach can be transferred to other markets; a similar project was carried out for the Swiss market.
- Price estimation for a property based on its parameters, such as house features, amount of land, available facilities, region and location
- Contribution of each property parameter to the overall price
- Market movement prediction that captures temporal dynamics and sends notifications on significant changes
- Automatic identification of underpriced properties on the market
- Interactive visualisation tool for the model’s insights
The outcome
The system lets the customer react quickly to market changes and gives it high-quality insights for potential investments, including market insights hidden from other stakeholders, which gave it a distinct competitive advantage.
Based on these insights, the customer made around €20 million of investments across multiple properties, with a clearly positive return on investment.
