需求預測對於國道客運經學者日常管理工作非常重要，可作為不同目標市場服務定價的參考依據。透過分析國內一家國道客運公司歷史訂票資料，本研究構建並比較迴歸、k個鄰近樣本法及加法型增量法模式在尖峰兩小時的預測績效並歸納出使用時機。其次，進一步探討如何透過敘述性偏好模式來量測在不同票種購買限制（出發時間、訂票時間、付款時間及退款比率）組合下使用者的願付價格。本文最終考量不同票種限制的組合以及折扣幅度，設計出數種符合市場旅客偏好的票種。情境試算結果發現，在需求彈性較大的出發時段實施多元化票種結構，將有助於提升營收。本研究探討需求預測及多元票種設計，研究成果可作為國道客運公司在擬定需求管理策略時的參考依據。

Demand forecasting provides essential information for intercity bus operators to conduct the fare menu aiming to attract passengers from multiple segments. As a result, it is critical for the bus corporation to be capable of integrating forecasting and pricing activities in the daily operation. This study focuses on analyzing real ticket sales data of a domestic bus company for two objectives. First of all, we construct regression, k nearest neighbor, and additive pick-up models for evaluating predictive performance and inducing the rule while applying models. Moreover, this study discusses on how to obtain willingness-to-pay of various ticket types considering different combinations of purchasing restrictions (fences) such as departure time, booking time, pay time, and refund by using the stated-preference model. We ultimately provide multiple customer-oriented ticket types given different combinations of restrictions and corresponding discounts. The results show that deploying the multi-fare structure during the period with high demand elasticity is helpful for increasing revenues. This study considers the integration of demand forecasting and fare design; the outcome may provide useful information for the intercity bus corporation to implement demand management strategies and increase revenues.