Exploring housing submarkets using time series clustering analysis: A case of the Seoul metropolitan area, South Korea

Keywords: housing submarkets, time series clustering analysis, sales price index, DTW, K-means, HDBSCAN
Abstract Type: Paper Abstract

Abstract

The housing market in South Korea has experienced unprecedented volatility with severe price fluctuations in the 2020s. Notably, this price variability of the real estate shows different patterns throughout regions because of various characteristics of housing submarkets. This study explores spatiotemporal characteristics of the house sales price within the Seoul metropolitan area, South Korea. Specifically, we identify the housing submarkets based on time series clustering methods such as K-means and HDBSCAN(Hierarchical Density-Based Spatial Clustering of Applications with Noise) using the DTW(Dynamic Time Warping) distance. As an empirical application, the sales price index of the apartment house is collected from 2014 to 2024 at the district level of the Seoul metropolitan area. For time series clustering, we generate hundreds of graphs for the sequences of the sales price index over time utilizing the SOM(Self-Organizing Map), an unsupervised learning method to map multidimensional variables onto a two-dimensional space. It is observed that the apartment sales price has continuously increased since 2014, with a dramatic rise in 2020, followed by a sharp decline in 2022. However, the spatiotemporal patterns of sales price fluctuations varied by region. As a result, five distinct housing submarkets were identified, and they demonstrated unique temporal patterns of the sales price variation: stepwise variable cluster, short-term variable cluster, stabilized cluster, underdeveloped peripheral cluster, and continuously rising cluster. Our findings would provide useful insights for understanding the housing market structure with high variability in South Korea, and contribute to establishing region-specific housing policies by distinguishing the submarkets.