ALF-Score-A novel approach to build a predictive network-based walkability scoring system

Area of Science:

• Urban Planning
• Public Health
• Geographic Information Systems (GIS)

Background:

• Walkability assessments are crucial for public health and physical activity promotion.
• Existing walkability measures often lack spatial resolution and do not incorporate user perspectives.
• Road network structures are fundamental to mobility but are underutilized in current walkability assessments.

Purpose of the Study:

• To develop a novel network-based walkability measure that integrates road network structures and user opinions.
• To create a high-resolution walkability scoring system applicable to any road segment.
• To improve upon existing area-based walkability measures by incorporating individual user perspectives.

Main Methods:

• Developed the Active Living Feature Score (ALF-Score), a network-based approach.
• Utilized machine learning models trained on road network features, road embeddings, and points of interest.
• Incorporated user opinions to establish a high-confidence ground-truth dataset for model training.

Main Results:

• The ALF-Score model achieved over 87% accuracy in estimating walkability.
• A conversion consistency of over 98% was maintained during model training.
• The proposed method demonstrated superior performance compared to existing walkability measures.

Conclusions:

• The ALF-Score provides a highly accurate and spatially resolved walkability assessment.
• This novel approach effectively integrates objective network data with subjective user opinions.
• The ALF-Score offers a valuable tool for urban planning and public health research, enabling personalized walkability evaluations.