Application of Machine Learning Classifiers for Mode Choice Modeling for Movement-Challenged Persons

In this study, we aimed to evaluate the performance of various machine learning (ML) classifiers predict mode choice movement-challenged persons (MCPs) based on data collected through a questionnaire survey 384 respondents in Dhaka, Bangladesh. The set consisted CNG-driven auto-rickshaw, bus, walking, motorized rickshaw, and non-motorized which was found as most prominent used by MCPs. Age, sex, income, travel time, supporting instrument (as an indicator level disability) utilized MCPs were explored predictive variables. Results from different split ratios with 10-fold cross-validation compared model outcomes. A ratio 60% demonstrates optimum accuracy. It that Multi-nominal Logistic Regression (MNL), K-Nearest Neighbors (KNN), Linear Discriminant Analysis (LDA) show higher accuracy for 60%. Overfitting bus walking source classification error. Travel time identified important factor influencing selection CNG, rickshaw MNL, KNN, LDA. LDA KNN depict more than MNL. follows relatively normal probability distribution, while distribution is negatively skewed other three modes.