Prediction and layout optimization of older adult(s) care facilities based on POI data, machine learning, and space...

Area of Science:

• Urban Planning
• Gerontology
• Data Science

Background:

• Rapid population aging presents significant challenges for urban planning and elder care facility placement.
• Traditional planning methods lack the spatial precision and practicality needed for effective elder care facility location.

Purpose of the Study:

• To develop and evaluate a three-tiered, progressive framework for predicting and optimizing the location of older adult care facilities.
• To integrate Point of Interest (POI) data, machine learning, and space syntax for precise spatial planning.

Main Methods:

• Divided Shanghai into 500x500m grid cells for analysis.
• Employed the C5.0 decision tree algorithm, benchmarking against logistic regression and random forests.
• Utilized a three-tiered screening: citywide prediction, aging data-based secondary screening, and space syntax verification.

Main Results:

• The C5.0 model achieved 83.36% accuracy, outperforming other algorithms.
• Identified 940 preliminary suitable grids, with key influencing factors including catering, real estate, and company facilities.
• Further screening identified 519 priority deployment grids and 104 key planning grids, with a case study highlighting Jiading District.

Conclusions:

• The developed framework effectively integrates macro-level suitability with micro-level accessibility for elder care facility planning.
• Provides actionable spatial decision-making references for urban planning authorities in high-density, aging cities.
• Offers a replicable methodological approach for planning public service facilities in similar urban contexts.