Rule-Based Screening Tool Ranks School Complex Facility Priorities Across South Korea

South Korea’s school-age population has fallen to 5.25 million in 2024, a decline of roughly 5.52 million since 1980, while the country’s total fertility rate dropped from 4.53 in 1970 to 0.68 in 2024. As classrooms empty and small schools with fewer than 240 students multiply, the Ministry of Education has since 2014 restricted the opening of new schools and adjusted the stock of existing ones, including closures. Since the first school closure in 1970, 3,955 schools have closed as of 2024. Most of those sites, 2,609, have been sold through public bidding, 979 have been repurposed through leases or direct government use, and 367 buildings remain unused.

Against this backdrop, researchers have developed a data-informed, rule-based decision-support framework for the preliminary planning of school complex facilities, the shared-use installations that integrate educational, cultural, welfare, sports, and parking functions within school premises. The framework, described in a study published in Architectural Research, does not attempt to predict actual facility demand. Instead, it combines administrative-unit-level demographic indicators with school-centered information on surrounding facilities to compare and screen potentially suitable facility types during the earliest planning stage, before site investigations, feasibility assessments, expert judgment, or stakeholder consultation take place.

The analytical logic is deliberately transparent. Facility types already present within the school itself, or those meeting a predefined local-duplication threshold of three or more instances within a 500-meter catchment, are excluded from the preliminary candidate set. Facility types associated with comparatively high demographic need and limited surrounding provision are given higher priority. The demographic assessment relies on resident registration population statistics from 2021 to 2024, categorizing residents into children and adolescents aged 0 to 18, working-age adults aged 19 to 64, and people aged 65 and above. The working-age share is shown only as context and does not enter the prioritization rules.

Rather than tracking absolute population change, the framework emphasizes each district’s relative proportions compared with the average of the corresponding upper-level administrative unit, such as a metropolitan city or province. When both the school-age and elderly population ratios of a district exceed the municipal average, both groups are treated as having relatively high potential welfare need. When only one group exceeds the average, the corresponding facility type, such as child welfare centers and care classes for a young population or senior welfare centers for an aging one, receives higher initial priority. When both ratios fall below the municipal average, relative proportions and rates of population change are evaluated independently, on the assumption that a higher relative share or a slower decline may be associated with greater potential demand.

The spatial component draws on Korean urban park planning standards, which prescribe catchment distances of 500 meters or less for neighborhood parks and 1,000 meters or less for walkable-neighborhood parks. An empirical Korean study found that walking 500 meters takes about 8.9 minutes for adults under 65 and about 14.8 minutes for adults 65 and older; the corresponding 1,000-meter times were extrapolated to roughly 17.9 and 29.6 minutes. The researchers therefore operationalized 500 meters as an immediate walking catchment of about 15 minutes and 1,000 meters as an extended catchment of about 30 minutes, while stressing that these are operational analytical thresholds, not universal service standards, and that actual walking times vary with age, physical condition, and speed.

The catchments are defined as circular radial buffers centered on each school rather than pedestrian-network service areas, meaning the analysis does not account for actual walking routes, street configuration, crossings, slopes, signal delays, or physical barriers. Facility counts within the 1,000-meter ring serve only as contextual information, while the 500-meter counts drive the local duplication screening and comparative prioritization. Because no established policy or empirical standard defines facility duplication, the count of three or more same-type facilities within 500 meters was adopted as an operational screening parameter, and the authors caution it should not be read as evidence of actual oversupply, since the framework does not incorporate facility capacity, scale, utilization rates, or service quality.

The framework was implemented in Python with a Streamlit web interface. Users select a school level, a provincial or metropolitan office of education, and a target school, or enter an address manually. The application then presents school characteristics such as teacher numbers and the presence of a library, visualizes age-specific population structures, and queries living SOC facilities within the two catchments through the Google Maps API, displaying them on an interactive Folium map with a summary table of counts by distance. The lower section of the interface lists prioritized facility types alongside excluded types and the reasons for exclusion, making the operation of the predefined rules visible to users and reducing the fragmented manual information gathering that currently characterizes project applications.

To test applicability, the framework was applied to all 81 school complex facility projects designated in 2023 and 2024, a full-sample exercise intended to demonstrate whether the rules could be implemented consistently rather than to validate the accuracy of the recommendations. Six cases were then selected for detailed illustration: two urban schools in Dobong-gu and Nowon-gu, Seoul; two rural schools in Uljin-gun, Gyeongsangbuk-do, and Jincheon-gun, Chungcheongbuk-do; and two closed-school sites in Yeoncheon-gun, Gyeonggi-do, and Gyeongsan-si, Gyeongsangbuk-do. In Dobong-gu, the elderly population share rose 3.9 percentage points from 2021 to 2024 while the 0-to-18 share fell 1.3 points, and senior welfare facilities emerged as the highest-priority option. Nowon-gu, with a modest 0.7-point elderly increase and a 1.5-point school-age increase, saw child welfare facilities and care classes prioritized instead.

The rural and closed-school cases followed the same pattern of divergence. Uljin-gun recorded a 6.7-point rise in the elderly share against a 2.4-point school-age decline, placing senior welfare facilities at the top, whereas Jincheon-gun’s relative increase in the school-age population led to child welfare facilities being selected as the final welfare recommendation. Yeoncheon-gun’s 14.3-point elderly increase and 5.0-point school-age decline classified elderly-oriented facilities as highest priority, while Gyeongsan-si, with a 4.6-point elderly decrease and a 0.4-point school-age increase, ranked child-oriented facilities higher. Even within the same locational categories, the framework generated different sets of facility priorities, reflecting differences in demographic trends, relative age-group proportions, and surrounding infrastructure.

The authors frame the contribution as a transparent and systematic basis for preliminary screening and comparative prioritization at the school level, supplementing uniform, district-based planning criteria with school-centered assessments of surrounding facility conditions. They are explicit about the limits: the outputs are conditional priorities generated under predefined rules, not validated assessments of actual facility demand, and the three-facility duplication threshold has not been empirically validated. Surrounding-facility data retrieved through the Google Maps API may vary with search keywords, categories, and update timing, and publicly disclosed school data may not reflect the latest facility conditions or utilization rates. Future work, they write, should test the robustness of the threshold with alternative parameters, incorporate facility capacity, utilization, and service quality, add dimensions such as disaster vulnerability and post-implementation evaluation, and validate outputs through field surveys and stakeholder engagement, with on-site investigation and expert consultation remaining essential for practical decisions.

Subject of Research: A data-informed, rule-based decision-support framework for preliminary screening and prioritization of school complex facility types in South Korea

Article Title: A Data-Informed, Rule-Based Decision-Support Framework for School Complex Facility Planning: Integrating Demographics, Surrounding Facilities, and Local Context

Article References: Jeong, J., & Yoo, H. (2026). A Data-Informed, Rule-Based Decision-Support Framework for School Complex Facility Planning: Integrating Demographics, Surrounding Facilities, and Local Context. Architectural Research, 28(1), Article 18. https://doi.org/10.1007/s44515-026-00020-2

Image Credits: AI Generated

DOI: 10.1007/s44515-026-00020-2

Keywords: school complex facilities, South Korea, decision-support framework, demographic change, school closures, facility planning, spatial catchment, welfare facilities, rule-based screening, population aging, GIS buffers, preliminary planning