Predictive Modeling of Multi-Channel Bus Ticket Demand With Price Sensitivity Analysis

Abstract

This study develops a predictive modeling framework for multi-channel bus ticket demand using counter ticket sales and online sales data from an aggregator-based e-ticketing system in Malaysia. The study integrates historical sales data with operational and external variables, including trip frequency, vehicle registration, public and school holidays, weather indicators, and ridership trends. A SARIMAX-based time series approach is applied together with comparative machine learning models to examine seasonal demand patterns, channel-level differences, and price sensitivity. The findings show that counter ticket sales and online sales exhibit different demand behaviors. Counter sales are relatively less sensitive to price changes and are more influenced by trip availability and seasonal travel periods, while online sales are shaped more strongly by operational efficiency, service availability, and customer purchasing behavior. The main contribution of this study is the use of price elasticity impact as a predictive variable to improve the interpretability of demand forecasting in a multi-channel ticketing environment. The proposed framework provides practical insights for bus operators to support pricing decisions, trip planning, and data-driven operational strategies in public transportation ticketing systems.