Behavioural Analysis of Urban Travel Mode Selection Using Adaptive Waterwheel Plant Optimized Random Forest (AWPORF) Algorithm

Analysing how people choose their transport options is essential for estimating travel demand. In addition to being recommended for modelling mode choice patterns, machine learning (ML) approaches are said to be useful for forecasting achievement. However, due to ML's black-box structure, it is tough to create a good explanation for the relationship between inputs and outputs. Using a novel Adaptive Waterwheel Plant Optimised Random Forest (AWPO-RF) method to analyse trip mode options, this research investigates the mathematical framework's predictability and interpretability. Applying the AWPO method improves the RF's prediction performance. Key metrics, including Mean Absolute Percentage Error (MAPE) and runtime, were used to evaluate the model. By optimizing the performance of the RF model, the AWPO-RF approach improves prediction accuracy in trip mode selection, attaining a 98.4% improvement in accuracy over conventional techniques. Furthermore, by predicting the weightings of the variables impacting mode choice, it improves interpretability and delivers insightful information on travel behaviour. Furthermore, the weightings of explicating factors are estimated using the AWPO-RF approach in regard to their connections with mode selections. This was crucial for comprehending and accurately simulating travel behaviours