Analysing house price dynamics using novel spatio-temporal methods
Guide(s)
Deb, Soudeep
Department
Decision Sciences
Area
Decision Sciences
University
Indian Institute of Management Bangalore
Place
Bangalore
Publication Date
3-31-2025
Year Awarded
March 2025
Year Completed
March 2025
Year Registered
June 2020
Abstract
In this thesis, three different problems related to house price dynamics are discussed. The first problem explores the application of a spatio-temporal model with a space-time error process and different residual distributions to analyze house price fluctuations in Bangalore, an emerging real estate market. The aim is to investigate how house prices change over time and space, considering the influence of factors such as local amenities. The analysis also examines whether different residual distributions improve model fit. The model accommodates multiple observations at the same location and time point, moving beyond the conventional approach of analyzing single observations for specific spatial and temporal units. A Bayesian framework is used, and Bangalore house price data serve as the case study. Building on this foundation, the second problem explores the space and time varying relationships between house prices and key amenities. Using insights from the first chapter, the model adopts a white noise error process and proceeds in two steps: identifying significant variables via a random forest algorithm and incorporating them into a spatial model fitted quarterly to capture temporal and spatial dynamics. Focusing on location and under construction metro stations as spatially varying regressors, the study highlights how nearby upcoming metro stations influence house prices through localized effects. This approach uncovers nuanced spatial patterns, providing deeper insights into Bangalore’s residential property market. The third problem tackles the computational challenges posed by large datasets, which limit the feasibility of the previous methods. A divide-and-conquer spatio-temporal modeling technique is introduced, partitioning spatial units into subsets, applying the model to each subset, and then combining the results. This method ensures accurate parameter estimation while reducing computational load. The approach is demonstrated using house price data from greater London, with a focus on handling missing data, illustrating its effectiveness for large-scale analysis. As a summary, this thesis addresses several intriguing problems within the real estate market and explores various approaches for addressing these issues through spatiotemporal modeling.
Pagination
xi, 128p.
Copyright
Indian Institute of Management Bangalore
Document Type
Dissertation
DAC Chairperson
Deb, Soudeep
DAC Members
Panchapagesan, Venkatesh; Dasgupta, Kunal; Deo, Anand
Type of Degree
Ph.D.
Recommended Citation
Gupta, Kapil, "Analysing house price dynamics using novel spatio-temporal methods" (2025). Doctoral Dissertations. 92.
https://research.iimb.ac.in/doc_dissertations/92
Relation
DIS-IIMB-FPM-P25-07
