Santi Phithakkitnukoon

Traditional pedestrian routing algorithms typically minimize physical distance or travel time, often overlooking contextual factors that influence route choice in digitally connected environments. As public WiFi infrastructure becomes increasingly prevalent in smart-city districts and university campuses, digital connectivity may influence pedestrian mobility decisions. This study introduces P-WARP, a multi-factor routing and inference framework that reconstructs latent pedestrian preferences by integrating physical effort, environmental walkability, and WiFi connectivity within a unified semantic graph. The empirical analysis is conducted on the Chiang Mai University campus, a digitally connected environment serving as a smart campus testbed. The framework integrates heterogeneous spatial datasets, including OpenStreetMap topology, Shuttle Radar Topography Mission elevation data, environmental walkability grids, and WiFi roaming logs collected via a custom mobile sensing application from 21 volunteers across 71 verified walking trips. Two routing strategies are evaluated: a Global Static Model, representing infrastructure-based connectivity assumptions, and a Trip-Centric Dynamic Model, incorporating realized connectivity histories. Model parameters are calibrated using Bayesian Optimization with five-fold cross-validation. Results show that incorporating realized connectivity reduces trajectory reconstruction error by 6.84% relative to the baseline. The learned parameters reveal a notable detour tolerance, suggesting that stable digital connectivity can influence pedestrian route choice in digitally instrumented environments.

Purpose: This study examines how spatial factors shape the popularity of coffee shops in Chiang Mai, Thailand and a mid-sized Southeast Asian city with a rapidly expanding café culture. It investigates whether proximity to urban amenities and neighboring cafés influences customer engagement reflected in online reviews and ratings. Design/methodology/approach: Data were collected in June 2025 using the Google Places API, identifying 2,468 coffee shops. Each entry was enriched with OpenStreetMap (OSM) data on nearby tourist attractions, universities, coworking spaces and transport hubs. A composite popularity score was constructed from ratings and review counts. Spatial clustering methods (DBSCAN, Moran's I and LISA) and correlation analysis were used to assess how popularity relates to locational characteristics. Findings: Popular cafés cluster in central districts, especially tourism areas, while proximity to general amenities shows weak associations with popularity. Newer or lower-performing cafés benefit from locating near successful peers, suggesting visibility spillovers. In contrast, top performers often succeed outside clusters, indicating differentiation or destination appeal. Practical implications: New entrants may enhance visibility by co-locating near established cafés, while mature businesses can sustain engagement through distinctive positioning in lower-density areas. Policymakers may support café districts such as Nimmanhaemin and the Old City while encouraging balanced dispersal. Originality/value: This study integrates digital reputation metrics with geospatial data to advance understanding of spatial embeddedness in hospitality entrepreneurship. Combining Google Places and OSM demonstrates how spatial positioning and online visibility jointly shape outcomes.

The rapid growth of dockless electric scooter (e-scooter) sharing services has transformed short-distance urban mobility, offering convenience and sustainability benefits while amplifying challenges related to demand imbalance, fleet rebalancing, and spatial inequity. Accurate spatiotemporal demand prediction is therefore essential for optimizing resource allocation and supporting data-driven policy interventions. This study proposes a hybrid deep learning framework that integrates a Graph Convolutional Network (GCN) with a Gated Recurrent Unit (GRU) and community detection to enhance short-term prediction of e-scooter pick-up and drop-off demands. The Louvain algorithm is employed to partition urban areas into mobility-based communities, enabling the model to capture functional connectivity rather than relying solely on geographic proximity. Using real-world e-scooter trip data from Calgary, Canada, the model’s performance is evaluated against established baselines, including a Masked Fully Convolutional Network (MFCN) and conventional GRU architectures. Results show that the proposed approach achieves up to 11.8% improvement in mean absolute error (MAE) compared with the MFCN baseline and more robust generalization across temporal horizons. The findings demonstrate that integrating community structures into graph-based learning effectively captures complex urban dynamics, providing practical insights for sustainable micromobility operation and service deployment.

This paper presents an initial exploration into understanding the intricate relationship between space utilization and electricity consumption within buildings. Leveraging Wi-Fi probe data for physical space utilization in 10 buildings at the Faculty of Engineering, Chiang Mai University, Thailand, the study introduces a novel approach to comprehend space utilization. Through the analysis of factors inferred from the Wi-Fi data including crowdedness, mobility, and entropy, correlated with changes in electricity consumption, the research aims to reveal strategies for optimizing real-time energy use. The temporal relationship is scrutinized by examining correlations between each space utilization attribute and electricity consumption at various time lags. Fluctuations in space utilization levels offer valuable insights into anticipated energy consumption in the near future. Results, in general, indicate that the 30–45 minutes timeframe serves as an early indicator for anticipated increases in electricity usage when considering crowdedness, while the 15–30 minutes timeframe is indicative when mobility and entropy are concerned. Buildings were clustered together based on electricity usage profiles and space allocation, yielding intuitive results. By delving into the dynamic interplay between space utilization and electricity consumption, the study contributes valuable insights to discussions on sustainable building practices and energy management.

Emerging on-demand shared mobility services face the difficulty of effectively balancing demand. Influx of these mobility services urges for more precise prediction of origin-destination demand becomes essential and urgent. Our previous work addressed this issue with a Masked Fully Convolutional Network (MFCN) model for short-term pick-up/drop-off demand prediction. In this study, we present a predictive modeling framework designed for short-term origin-destination demand prediction. This framework harnesses the capabilities of Convolutional Neural Networks (CNNs), integrates our previously developed MFCN model, and introduces novel prediction fusion and scaling methodologies. Furthermore, a new loss function is developed and designed to effectively train the model with demand and location information. We evaluated the proposed framework using shared e-scooter trip data from Calgary, Canada. Our evaluation encompasses two prediction scenarios: next-hour and next-24-hour predictions. The performance of our framework is benchmarked against baseline models including the naïve predictor, linear regression, GCN, and variant models. Our model shows the best performance regarding the true positive and F1-score values. The results suggest a high degree of regularity in the daily demand as the next-24-hour predictor performs better than the other scheme. Nonetheless, when a spatial error is considered, the performances of the two prediction schemes are comparable.

This work introduces an alternative solution to costly conventional approaches for large-scale travel behavior data collection by utilizing an opportunistic sensing data source i.e., Wi-Fi probe data. Through our case study of Chiang Mai University campus as a city, we developed a framework for inferring and visualizing Wi-Fi data-based travel behavior by demonstrating how a Wi-Fi probe data can be analyzed to infer trips and origin-destination flows. Specifically, our contributions include algorithms developed for inferring spatial presence, residence, stay, trip, and trip distribution among places in the campus, as well as campus inflow and outflow. Moreover, to handle the Wi-Fi access point data for the analysis, and visualize the inferred trips and flows, an online visual analytics tool called Wi-Flow is developed as part of this work. Our framework differs from the other studies with our residence and trip detection algorithms that produce the result at the individual level as opposed to the overall network. The experimental results are intuitive and insightful, providing useful information for area management. Our work highlights the usefulness of Wi-Fi probe data in mobility modeling, and, in general, paves the way for opportunistic sensing approach to estimating mobility flows.