Przegląd Geograficzny (2026) vol. 98, iss. 2
A Typology of Housing in Polish Municipalities Using Gaussian Mixture Models Algorithm
Przegląd Geograficzny (2026) vol. 98, iss. 2, pp. 179-204 | 
Full text
doi: https://doi.org/10.7163/PrzG.2026.2.3
Abstract
This study addresses a significant gap in research on housing in Poland. Previous analyses have typically focused on selected aspects of housing, such as living conditions, demographic change, or spatial variation in housing quality. However, comprehensive, data-driven approaches integrating demographic pressure, the dynamics and structure of new housing supply, and the characteristics of the existing housing stock remain scarce.
The main objective of this study is to propose a typology of municipalities in Poland based on a multidimensional analysis of housing-related data. An exploratory approach was adopted, using a clustering method based on the Gaussian Mixture Models (GMM) algorithm. Importantly, the model was not constrained to produce predefined clusters; instead, theoretically expected types and subtypes, such as a metropolitan urban subtype, other urban subtypes, and a ʻrecreationalʼ type characterized by a high share of seasonal housing, were treated as working hypotheses subject to empirical verification.
As a result, five main types of housing systems were identified: urban, suburban, intermediate, peripheral, and special types. These differ in terms of demographic dynamics, the structure of investors, the intensity of housing production, and the condition of the housing stock. Each type was further divided into subtypes reflecting distinct trajectories of housing development and reproduction, yielding a total of fifteen subtypes. To illustrate differences between types, a typogram method was used, based on percentile distributions of key variables.
The distribution of municipalities and housing stock across types reveals substantial structural disparities. Out of 2,477 municipalities, only five (0.2%) were classified as metropolitan urban units, yet they accounted for as much as 14.5% of the national housing stock in 2024. More broadly, urban type comprised 8.0% of municipalities but concentrated 50.9% of housing resources. In contrast, suburban areas included 27.2% of municipalities but only 18.4% of the housing stock. A particularly notable imbalance is observed in dispersed suburban areas, which account for 11.9% of municipalities but just 5.6% of housing stock. Peripheral areas exhibit the opposite pattern: they encompass 48.5% of municipalities but contain only 15.6% of the national housing stock. This highlights the fragmented and relatively underdeveloped nature of housing stock in these regions. Exceptions to this pattern include compact suburban zones, typically surrounding large cities, and recreational type, both of which have a slightly higher share of housing stock relative to their share of municipalities.
The spatial distribution of the identified types reveals clear regional patterns. Urban and suburban types are relatively evenly distributed across the country, reflecting the nationwide presence of cities and their surrounding zones. In contrast, peripheral types display a strong east-west differentiation. In eastern and central Poland, peripheral areas are predominantly characterized by deprivation, reflected in insufficient housing infrastructure and lower-quality stock. In western and northern regions, peripheral areas are more often defined by obsolescence, with a higher share of aging housing stock.
The recreational type is concentrated in areas traditionally associated with seasonal tourism, including coastal zones, lake districts, and mountainous regions. It is also present in the vicinity of larger urban agglomerations, where it includes spa towns and municipalities with a high share of newly built, well-equipped, and low-density housing. A separate public type is more dispersed but occurs most frequently in northern Poland.
At the level of subtypes, further spatial differentiation becomes evident. Metropolitan urban municipalities are limited to five major cities with highly dynamic, market-driven housing systems. Other urban subtypes include stable and stagnating cities and towns, often surrounded by distinct suburban zones, compact in larger agglomerations, and more dispersed in smaller ones. Peripheral subtypes include both deprivation-related and obsolescence-related forms, with specific spatial concentrations, such as deficit-type deprivation in southeastern Poland and along the northern boundary of its occurrence, and deep housing obsolescence in the Sudetes. Intermediate subtypes show patterns broadly similar to peripheral ones, with stagnation dominating in central and eastern regions and obsolescence in the west and north.
The findings have both practical and scientific relevance. From a practical standpoint, they may contribute to analyses concerning spatial planning and the identification of housing policy needs across territorial administrative units. From a research standpoint, the study demonstrates the value of integrating demographic, supply-side, and stock-related variables in the analysis of housing. It also shows that housing processes can be effectively understood through the relationships between population dynamics, housing production, and existing stock, rather than solely through market conditions or spatial morphology.
At the same time, the study has limitations. The analysis was conducted at the municipal level, while migration data were only available at the county level, limiting the ability to capture fine-grained spatial variation in demographic processes. Future research could focus on refining the typology, particularly within heterogeneous categories such as stable and stagnating urban types, which include municipalities with diverse functional roles and positions within the national settlement system.
Keywords: housing, typology of municipalities, Gaussian Mixture Models, demographic demand, housing production, housing stock.
[hubert.horynek@pw.edu.pl], Politechnika Warszawska, Wydział Geodezji i Kartografii
Citation
APA: Horynek, H. (2026). Typologia mieszkalnictwa w gminach Polski z wykorzystaniem algorytmu Gaussian Mixture Models. Przegląd Geograficzny, 98(2), 179-204. https://doi.org/10.7163/PrzG.2026.2.3
MLA: Horynek, Hubert. "Typologia mieszkalnictwa w gminach Polski z wykorzystaniem algorytmu Gaussian Mixture Models". Przegląd Geograficzny, vol. 98, no. 2, 2026, pp. 179-204. https://doi.org/10.7163/PrzG.2026.2.3
Chicago: Horynek, Hubert. "Typologia mieszkalnictwa w gminach Polski z wykorzystaniem algorytmu Gaussian Mixture Models". Przegląd Geograficzny 98, no. 2 (2026): 179-204. https://doi.org/10.7163/PrzG.2026.2.3
Harvard: Horynek, H. 2026. "Typologia mieszkalnictwa w gminach Polski z wykorzystaniem algorytmu Gaussian Mixture Models". Przegląd Geograficzny, vol. 98, no. 2, pp. 179-204. https://doi.org/10.7163/PrzG.2026.2.3