2026-07-29

Ordinal regression and explainable machine learning reveal key determinants of urban bus service quality in Dhaka

This study evaluated urban bus service quality in Dhaka using a SERVQUAL questionnaire survey administered to 500 passengers across five major bus routes. Researchers modeled passenger satisfaction and dissatisfaction using Ordinal Logistic Regression alongside Random Forest and Extreme Gradient Boosting classifiers, achieving classification accuracies of 0.71 and 0.70 respectively. SHAP analysis identified that the largest service gaps and drivers of dissatisfaction relate to safety concerns, unprofessional staff behavior, scheduling, and poor vehicle maintenance.

Why it matters — It identifies the specific psychological and operational dimensions—specifically assurance and empathy—that drive passenger dissatisfaction, moving beyond general service ratings to show transit operators exactly which soft and hard service elements must be corrected to halt the shift toward private vehicles.

Caveat: The findings are based on a relatively small sample of 500 respondents restricted to five specific routes in a single city, which may not fully represent the diverse transit experiences across the entire metropolitan area or other developing urban centers.

Discover Cities · pdf · doi

Cultural Care? The European capitals of culture as ambivalent infrastructures of urban (Un)care. The cases of Matera and Plovdiv 2019

This study evaluates the 2019 European Capitals of Culture (ECoC) programmes in Matera, Italy, and Plovdiv, Bulgaria, using a qualitative, comparative framework to analyze how these large-scale cultural events affect urban social cohesion. The research traces the tensions between participatory rhetoric and actual implementation, finding that while both cities successfully mobilized resources and gained visibility, they suffered from centralized governance, short-lived public participation, and weak long-term institutional support.

Why it matters — It demonstrates that culture-led urban development initiatives do not automatically foster social inclusion or community care; instead, they function as ambivalent infrastructures where participatory goals are frequently undermined by top-down governance structures.

Caveat: The findings are based on qualitative case studies of only two specific European cities during their 2019 designation year, which may limit generalizability to other ECoC hosts with different institutional contexts.

City Culture and Society · doi

“Food demons have taken over our cities”: Unpacking foodification and airbnbification of an historical market in a southern Europe city

This study analyzes the spatial and temporal links between foodification, touristification, and short-term rentals through an in-depth case study of a historical fish market area in Catania, Italy. Using a mixed-method approach that integrates on-site field surveys, GIS mapping, and long-term ethnography, the research tracks how food-led commercial restructuring transitions into tourism-driven displacement under weak local governance.

Why it matters — It demonstrates that food-led retail changes act as an initial catalyst that actively co-produces and accelerates subsequent touristification and Airbnb expansion, rather than these processes merely occurring in parallel.

Caveat: The findings are based on a single case study of a mid-sized Southern European city with highly fragmented local governance, which may limit generalizability to cities with stronger regulatory frameworks.

Cities · doi

Recreation versus conservation: The capitalization of parks and protected areas in the German housing market

Using a hedonic regression and difference-in-differences framework on German housing data from 2007 to 2023, this study measures how proximity to parks versus protected areas (PAs) affects residential property values. Proximity to parks increases home prices, yielding an estimated aggregate value of ‐983 billion (about 4% of the housing stock), which disproportionately benefits wealthier households due to higher baseline property values. Conversely, properties located within PAs experience substantial price discounts, resulting in a highly uncertain aggregate PA value estimated at -‐5.2 billion.

Why it matters — The study demonstrates that the capitalization of open space depends heavily on its primary function, showing that recreational parks command a premium while conservation-focused protected areas can actually depress property values for homes located within their boundaries.

Cities · doi

A tale of many cities: The role of inter-municipal food networks in changing the dominant food regime

This study conducts a systematic literature review of academic research on inter-municipal food networks (IFNs) to analyze their scale, governance structures, political activities, and operational barriers. Using a multilevel perspective framework, it synthesizes how these networks of local governments facilitate mutual exchange and policy advocacy to shape urban food strategies.

Why it matters — It shifts the research focus from isolated, single-city case studies to the collective dynamics of municipal networks, revealing how collaborative local governance can scale up to challenge dominant, unsustainable food regimes.

Caveat: The analysis is constrained by significant gaps in the existing literature regarding the precise organizational governance, scale, and measurable impacts of these networks, which limits a definitive assessment of their transformative power.

Cities · pdf · doi

Local governments, gender balance and service delivery performance

This study analyzed the relationship between gender balance in local governments and municipal service delivery performance using data envelopment analysis and regression models. The sample comprised 102 Spanish municipalities with populations over 50,000 during the 2022 fiscal year. The analysis revealed that greater gender balance within the municipal government team improves service delivery performance specifically when a female mayor is in office, with further performance gains observed under the interaction of right-wing female mayors and gender diversity.

Why it matters — It demonstrates that the performance benefits of gender diversity in local cabinets are contingent on leadership characteristics, showing that gender balance does not automatically improve efficiency unless paired with a female mayor.

Caveat: The findings are based on a single fiscal year (2022) across Spanish municipalities with more than 50,000 residents, which may limit generalizability to smaller towns or different national governance systems.

Cities · pdf · doi

Mapping thematic evolution in urban design for older people: A hybrid scientometric review

This study maps the thematic evolution of urban design for older people by analyzing 787 publications spanning 2016 to 2025. Using a hybrid analytical approach that combines S-curve life-cycle modeling, Latent Dirichlet Allocation topic modeling, and cosine similarity, the researchers traced the field's progression across four distinct stages and identified three primary evolutionary roadmaps: healthy ageing and smart environments, neighborhood and spatial equity, and policy and service provision.

Why it matters — It establishes a structured, computational framework that tracks how research has shifted from basic community planning toward smart, inclusive, and technology-integrated urban strategies, providing planners with an evidence-based roadmap of emerging design priorities.

Caveat: The analysis is restricted to the thematic trends present within the selected academic literature database and does not evaluate the real-world implementation or efficacy of these urban design strategies.

Habitat International · doi

Interventions to increase cycling: A critical assessment of effectiveness

This study synthesized 124 empirical evaluations of cycling interventions using a rule-based classification framework to compare outcomes across diverse study designs. It analyzed interventions by type, cost, and behavioral mechanisms, assessing the direction, strength, and confidence of their impacts on cycling behavior.

Why it matters — It demonstrates that structural changes like bike paths often merely redistribute existing riders rather than recruit new ones, and that behavioral campaigns rarely sustain change alone. Instead, interventions that directly improve practical access, such as e-bike subsidies and bike-share schemes, are the most consistent drivers of new cycling uptake.

Caveat: The synthesis relies on a heterogeneous pool of existing studies with varying designs and measures, which may limit direct comparability.

Journal of Cycling and Micromobility Research · pdf · doi

Accessibility as practice: Non-visible disability and lived health experiences in public transport

Using 50 semi-structured interviews and a three-month ethnography conducted between 2023 and 2025 in Hong Kong, this study analyzes how passengers with non-visible disabilities, such as post-cancer exhaustion and acquired brain injury, navigate the public transport system. The research details how social norms favoring visible impairments force these passengers to engage in health-compromising adaptations, including avoiding priority seats, rerouting to escape public scrutiny, or switching to more physically demanding transit modes.

Why it matters — It demonstrates that transit accessibility is not merely an infrastructural or proximity metric as framed by SDG 11.2.1, but a socially negotiated practice. This shifts the focus toward experiential usability, showing how formal transit systems fail when they rely on visible markers of disability to grant legitimacy.

Caveat: The findings are based on qualitative research and ethnography within a single, highly dense transit environment, which may limit direct generalizability to cities with different cultural norms around public space and social interaction.

Journal of Transport & Health · doi

Comparing waterbody cooling during heatwave with subsequent summers in high-latitude coastal Copenhagen

This study measured the land surface temperature cooling effects of coastal seas and lakes in Copenhagen across three June days in 2018, 2019, and 2020, capturing a dry heatwave, a windy warm day, and a cooler moist day. The coastal sea provided 3.92–4.91°C of cooling up to 195 meters away, while lakes larger than 4.68 hectares cooled their surroundings by 2.81–3.27°C up to 285 meters. Lakes surrounded by tall vegetation showed smaller direct cooling extents but had significantly lower surrounding land surface temperatures of -0.6 to 0.8°C.

Why it matters — It quantifies how the cooling performance of urban blue infrastructure changes under different meteorological conditions, demonstrating that while waterbody cooling intensity is lowest during windy conditions and constrained during dry heatwaves, combining larger waterbodies with tall vegetation offers a dual pathway for urban heat mitigation.

Caveat: The findings are based on observations from only three specific days, which may not capture the full seasonal or interannual variability of waterbody cooling dynamics.

Sustainable Cities and Society · doi

CAPITALIZING CRISIS: Urban Extractivism in Athens Amidst Neoliberal Restructurings

The study conceptualizes post-2009 Athens through the lens of urban extractivism, analyzing how the Greek sovereign debt crisis institutionalized extractive governance. Drawing on legislative analysis and four years of ethnographic research in rapidly transforming neighborhoods, it traces how urban land, housing, infrastructure, and public commons are treated as exhaustible resources for transnational capital.

Why it matters — It reframes Southern European cities not as passive sites of austerity adjustment, but as active, systematically reorganized territories designed for repeated cycles of urban value extraction through exceptional planning regimes and real estate financialization.

International Journal of Urban and Regional Research · doi

Neural Architecture Search for Traffic Prediction: A Survey of Methods, Challenges, and Future Directions

This survey reviews the application of Neural Architecture Search (NAS) to traffic prediction, categorizing existing approaches into gradient-based, evolutionary, and one-shot weight-sharing methods. It evaluates how these automated techniques design search spaces to capture spatial-temporal traffic patterns and balances computational costs against model quality.

Why it matters — It systematizes the transition from manually designed deep learning models, which require extensive expert labor and generalize poorly, to automated search strategies capable of discovering optimal architectures for diverse urban road networks.

arXiv · pdf

Structure of System-Specific Hazard Modeling for Physical Infrastructures in a Climate Change Context

The paper introduces the Boolean Hazard Model with Probability, a threshold-based framework designed to model climate hazards based on the specific vulnerabilities of physical infrastructure rather than generic environmental extremes. The model's utility is demonstrated through use cases that generate time-evolving hazard maps tailored to distinct infrastructure characteristics and hazard types.

Why it matters — It shifts hazard modeling away from generic regional climate extremes to system-specific thresholds, preventing planners from misallocating adaptation resources to climate shifts that do not actually threaten the physical integrity of specific infrastructure assets.

Caveat: The framework is demonstrated using a simplified threshold-based model, which may not capture complex, cascading, or non-linear physical interactions between climate stressors and infrastructure systems.

arXiv · pdf

Object Detection for Autonomous Driving in Chinese Rural Scenes: An Experimental Study on Real-Synthetic Data Mixing and Model Evaluation

The study evaluates 13 mainstream object detection models, including YOLOv5, YOLOv8, YOLO11, YOLO26, and RT-DETR-L, using a custom dataset of real-world images from Weishi County, Henan Province, mixed with synthetic images from Unreal Engine. The dataset features a 14-category classification system tailored to Chinese rural roads, capturing local elements like electric tricycles and low-speed vehicles. Testing across different real-to-synthetic ratios revealed that a 1:0.5 mix optimized performance, with YOLO11m achieving the highest mAP@0.5 of 0.758, whereas a 1:1 ratio degraded performance due to domain shift.

Why it matters — It establishes a benchmark and data-mixing strategy for rural autonomous driving, demonstrating that moderate synthetic data integration improves detection while identifying persistent failure modes in recognizing non-standard, long-tail objects like roadside stalls.

Caveat: The real-world baseline data is sourced from a single county in Henan Province, which may not capture the full diversity of rural road environments across other regions of China.

arXiv · pdf