An investigation into the determinants of satisfaction concerning varied toll policies on highways using the random forest model

Abstract

To address the insufficient consideration of public satisfaction's impact on the assessment of the implementation effectiveness of differentiated toll collection on expressways, a study was conducted using a satisfaction survey questionnaire from differentiated toll road sections in Yunnan Province in 2022. A random forest model (RF) was constructed under a two-category experiment to analyze the factors influencing satisfaction with expressway-differentiated toll policies. Multiple models underwent five-classification and two-classification experiments using the same training and test datasets. Results revealed that the RF model in the binary classification experiment exhibited a good fit. Notably, the satisfaction level with timely and accurate preferential policies emerged as the most critical factor, contributing 20.35% to the overall satisfaction with expressway differentiated toll policies. Independent effect analysis highlighted that the overall satisfaction with the differentiated charging method for empty trucks ranked highest, while satisfaction with the differentiated charging method for road sections was the lowest.

Keywords

• differentiated charges
• satisfaction evaluation
• influencing factors
• random forest model
• logit model