The travel mode preference exists in both culture and theenvironment. The wide scale of people's mobility makesour cities more polluted and congested, eventually affecting urban assets.Understanding people’s mode choice is important to develop urbantransportation planning policies effectively. This study aims to model andpredict the commuter’s mode choice behaviour in Lahore, Pakistan. A surveywas conducted, and the data was used for model validation. The comparative study was further done among multinomial logit model (MNL),Random Forest (RF), and K-Nearest Neighbor (KNN) classification approaches. It’s common in existing studies that vehicle ownership is rankedas the most important among all features impacting commuters’ travel modechoice. Since many commuters in Lahore own no vehicle, it’s unclear whatthe rank of factors impacting non-vehicle owners is. Other than thecomparison of predicting the performance of the methods, our contributionis to do more analysis of the rank of factors impacting the different types ofcommuters. It was observed that occupation is ranked as the most importantamong all features for non-vehicle owners.
Keywords : Travel Behavior, Machine Learning, Multinomial Logit Model, Random Forest, K-nearest Neighbor, the Travel Mode Choice
Authored by :
(1) Fariha Tariq
Department of City and Regional Planning, University of Management and Technology, Lahore, Punjab, Pakistan.
(2) Nabeel Shakeel
Department of City and Regional Planning, University of Management and Technology, Lahore, Punjab, Pakistan.
GSSR struggles for a future where human development is based on informed decision making process. Therefore, the journal challenges the status quo of current academic research that suffers from disconnect to actual application by finding solution to the problems confronted by humanity. GSSR intends to become the global leader in introduction of innovations where theories connect with application while maintaining the best features of theoretical and applied aspects.
