| dc.description.abstract | Can you afford to buy a place of your own? If you could, how would you go about finding it? Due to rising real estate prices in cities all over the world, more people than ever choose to rent in shared accommodations to save on costs. With whom and where we live has a significant impact on our lives, yet despite its importance, it is still one of the most manual and laborious processes in our digital age.
In this thesis, a prototype of a group recommendation system for real estate is presented to solve this issue. The prototype uses the context of where the users are looking to move, their habits and personality, as well as what they are looking for in a property, to cluster them into match groups. Then it provides housing recommendations for the group from various data sources based on the group's preferences. The research has been conducted within the Design Science Paradigm.
Presented in this thesis, is the design and implementation of the prototype in addition to the state of the art of both group and real estate recommendation. It provides a quantitative analysis of the different clustering algorithms, and qualitative evaluations of the user interface and real estate recommendations. The analyses of the matching algorithms show that a cascading hybrid model doing an initial clustering with k-Means, followed by a final grouping with a modified version of k Nearest Neighbour performs best for this problem. Further, the results show that users have a positive experience with the prototype, both with the interface and the real estate recommendations. Several topics for future research have also been identified. | |
