Reconceptualising transit ridership through locational utility: integrating OLS, GWR and ML in Kolkata’s MRTS

Urban growth has transformed cities, with global urban population rising from 34% in 1960 to 57% in 2022. Due to rapid urbanisation, challenges like traffic congestion and pollution have prompted investments in public transit systems as Mass Rapid Transit Systems (MRTS) and Bus Rapid Transit Systems (BRTS). This paper investigates the case of Kolkata MRTS in India to assess how “locational utility”—conceptualised as the combined influence of place-specific factors (density, land uses, walkability) and station-amenity factors affect ridership. Both global (Ordinary Least Squares-OLS) and local (Geographically Weighted Regression-GWR) models are used to test two analytical cases i.e., Case 1 using only place-specific variables and Case 2 using both place-specific and station-amenity variables. The adjusted R-squared of GWR, accounting for significant spatial variations and locational specificities, surpassed those of the corresponding OLS model in both cases. Further, adjusted R-squared improved substantially in GWR, explaining 79.2% of the variations in Case 1 to 92.7% in Case 2. Our models confirm on the role of land use, feeder transit stops, years of operation, comfort, safety and cleanliness in stations as robust predictors of transit ridership. Random Forest (RF) model was additionally employed as a robustness check to confirm relative importance of key predictors. The findings suggest importance of spatial heterogeneity and relevance of locational utility in shaping ridership patterns and offer practical insights for spatially informed urban transit planning.