The project addressed the real estate sector's need for fast, large-scale value estimates, useful as a preliminary reference in decision-making, in a field where traditional valuation requires the individual involvement of a certified professional.
The objectives of the project were:
- Develop a platform capable of carrying out mass property valuations automatically and without human assistance.
- Provide fast estimates close to the real market value, useful as an indicative reference.
- Integrate heterogeneous sources of information into the valuation, including official sources as well as data from the market and the Internet.
The areas around which TAKSON was structured were:
- Master system with artificial intelligence: algorithms that estimated the value of properties automatically and without human assistance, also making it possible to calibrate appraisals carried out by traditional methods against the automatic estimate.
- Integration of heterogeneous sources: ETL processes that combined real geopositioned comparables from the management systems of appraisal companies, comparables obtained from real estate portals, opinions about areas, neighborhoods, and municipalities extracted from social networks and online press, cadastral information, and statistical data from the INE. This approach set TAKSON apart from other online valuation solutions, which are based solely on published listings.
- Geopositioning and visualization: visual capabilities to locate properties based on information from the master system and cadastral data, with heat maps by zone type.
AESVAL and SERESCO relied on CTIC, whose contribution focused on developing the platform's artificial intelligence engine and the ETL processes for acquiring and integrating data from heterogeneous sources.
