The main goal of the project is to automatically identify similar geographical areas in terms of real estate market behaviour, socio-demographic and built-in characteristics. For this purpose, clustering algorithms (K-means, DBScan) are being used. For this purpose, the project will combine traditional geodemographic techniques with state of the art (unsupervised) machine learning and novel sources of data from Rightmove. The inclusion of some of these sources will also likely require to introduce methodologies able to extract information from image processing analysis.
Deadline 14th April