Prototyping automated property portfolio analysis and investment suggestions

Designing evaluation schemes for cluster analysis of high-dimensional data using biased example clusters
Students: Rasmus Friis Jensen
Supervisors: Niels Ørbæk Chemnitz, Philippe Bonnet
Level: MSc, Semester: Spring 2021
Tags: Unsupervised Learning, Recommender Systems, High-dimensional Data, Experimental Design

The industry of real estate investment is associated with great imbalance, since the demand for investments by far succeeds the supply. The process of identifying a potential new investment is prolonged and includes complicated searching. Data relevant for a property is spread across several sites and databases, and on top of that only a few portals offer the service of scanning the real estate market for properties not officially listed. All in all, these circumstances make it difficult to identify properties fitting well into an existing real estate portfolio.

The goal of the project is to identify and employ approaches for performing data wrangling on Danish real estate data. Moreover, the project aims to design, implement, and evaluate a prototype, that based on real estate data can perform property portfolio analysis and provide new investment suggestions by utilizing machine learning techniques.