國內整體社會的少子化趨勢、醫療體系的完善，帶動著高齡社會的到來。目前我國的被照顧需求者的比例逐年增加，政府因應長期照顧需求制定相關長照政策，於2017年實施的長期照顧十年計畫2.0中，為保障被照顧需求者之基本運輸需求，各縣市以復康巴士提供交通接送服務。
然而，乘客就醫行為使復康巴士需求具明顯尖離峰特性，業者在人工排班上面臨困境、供需不平衡的情況下，乘客需求無法被有效滿足、預約不成的比率高居不下，而現行提前預約的制度，無法實際符合臨時性的乘客需求。
近年來，國外運輸相關研究大量將預測技術與運輸領域結合，以預先的需求預測確保有效的運輸資源配置，目前國內復康巴士的學術研究多著重於派遣機制、路線規劃、滿意度分析等，鮮少以預測技術探討復康巴士需求議題。
本研究以花蓮縣的復康巴士為對象，透過五種機器學習方法建立辨別臨時需求之模型，協助提供預先保留運能的依據；使用近年較為主流、長時序預測上相對穩健的長短期記憶法，建構預測需求量模型，並嘗試辨別各種因素對復康巴士需求的影響程度，以利業者可於派遣前端作業進行提前運量規劃，降低預約不成的比率。
研究結果顯示，復康巴士的需求預測適合以其停放之場站區域進行預測分析：在預測臨時性需求時，高需求區域可採行邏輯迴歸建立預警模型，低需求區域則可使用集成學習方法，以預先得知具有8成以上準確率的判斷結果；在預測需求量時，外部因素影響程度小，僅需單純考量乘客本身需求即可，預測損失（MAE）降至0.09-0.12間。後續以研究結果與現行制度進行相關探討，提出管理建議予復康巴士業者參考。

The trend of declining birth rate in Taiwan as a whole and the improvement of the medical system have led to the arrival of an Aging Society. The elderly people and disabled in Taiwan is increasing year by year. In the Long-Term Care 2.0 (LTC2.0) implemented in 2017, ensure the rignts of people in need of care, each city provides transportation services with Rehabilitation Bus.
However, the Health seeking behavior of passengers makes the demand for Rehabilitation Buses have an obvious peak characteristic. The operators are facing difficulties in manual scheduling and the imbalance between supply and demand. The reservation system cannot actually meet the temporary passenger needs.
In recent years,foreign transportation research have combined forecasting with the transportation field to ensure effective transportation resource allocation with demand forecasting. At present, domestic research on Rehabilitation Bus mostly focuses on dispatching mechanisms, route planning, satisfaction analysis, etc. They seldom use forecasting to discuss the issue of demand for Rehabilitation Bus.
This study takes the Rehabilitation Bus in Hualien as the object, and uses five machine learning methods to establish a model for identifying temporary needs,help provide a basis for pre-reserving transport capacity; using the Long Short-Term Memory method to construct a forecast demand model, and try to identify the degree of influence of various factors.The operators can plan in advance at the front-end operation of the dispatch, and reduce the rate of unsuccessful reservations.
The results show that the demand forecast of Rehabilitation Bus is suitable for predictive analysis based on the area where they are parked: when predicting temporary demand, Logistic Regression can be used to establish an model for high-demand areas, and Ensemble Learning can be used for low-demand areas , with an accuracy rate of more than 80%; When forecasting demand, the degree of influence of external factors is small, and it is only necessary to simply consider the needs of passengers themselves, and the forecast loss (MAE) is reduced to 0.09-0.12. In the follow-up, relevant discussions will be conducted based on the results and the current system.Management suggestions will be put forward for the reference of Rehabilitation bus operators.

第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 7
1.3 研究範圍 7
1.4 研究流程 8
第二章 文獻回顧 11
2.1 各國復康巴士營運模式 11
2.1.1 加拿大 11
2.1.2 日本 13
2.1.3 香港 14
2.1.4 台灣 16
2.2 分類預測問題 19
2.3 需求預測問題 26
2.4 小結 32
第三章 研究方法 35
3.1 研究課題 35
3.1.1 課題一:預測目標區域隔日是否有臨時需求 35
3.1.2 課題二:預測目標區域隔日的總需求量 36
3.2 課題一 38
3.2.1 資料說明 38
3.2.2 資料處理與特徵工程 40
3.2.3 模型建立 47
3.2.4 評估指標 50
3.3 課題二 53
3.3.1 資料說明 53
3.3.2 外部因素、時間因素 58
3.3.3 預測模型 60
3.3.4 評估指標 68
第四章 案例分析 69
4.1 課題一 69
4.1.1 區域預測分析 69
4.1.2 鄉鎮預測分析 71
4.1.3 模型超參數配置 78
4.1.4 實作結果探討 80
4.2 課題二 84
4.2.1 模型成效 84
4.2.2 延續性探討 91
第五章 結論與建議 97
5.1 結論 97
5.1.1 課題一 97
5.1.2 課題二 100
5.1.3 小結 103
5.2 管理意涵 104
5.3 未來研究建議 108
參考文獻 111

Ai, Y., Li, Z., Gan, M., Zhang, Y., Yu, D., Chen, W., & Ju, Y. (2019). A deep learning approach on short-term spatiotemporal distribution forecasting of dockless bike-sharing system. Neural Computing and Applications, 31, 1665-1677.
Amrane, M., Oukid, S., Gagaoua, I., & Ensari, T. (2018). Breast cancer classification using machine learning. Paper presented at the 2018 electric electronics, computer science, biomedical engineerings' meeting (EBBT).
An, G., Omodaka, K., Tsuda, S., Shiga, Y., Takada, N., Kikawa, T., . . . Akiba, M. (2018). Comparison of machine-learning classification models for glaucoma management. Journal of healthcare engineering, 2018.
Askari, B., Le Quy, T., & Ntoutsi, E. (2020). Taxi demand prediction using an LSTM-based deep sequence model and points of interest. Paper presented at the 2020 IEEE 44th Annual Computers, Software, and Applications Conference (COMPSAC).
Bi, J.-W., Liu, Y., & Li, H. (2020). Daily tourism volume forecasting for tourist attractions. Annals of Tourism Research, 83, 102923.
Ceccarelli, G., Cantelmo, G., Nigro, M., & Antoniou, C. (2021). Machine Learning from imbalanced data-sets: an application to the bike-sharing inventory problem. Paper presented at the 2021 7th International Conference on Models and Technologies for Intelligent Transportation Systems (MT-ITS).
Chandakas, E. (2020). On demand forecasting of demand-responsive paratransit services with prior reservations. Transportation Research Part C: Emerging Technologies, 120, 102817.
De Caigny, A., Coussement, K., & De Bock, K. W. (2018). A new hybrid classification algorithm for customer churn prediction based on logistic regression and decision trees. European Journal of Operational Research, 269(2), 760-772.
Faghih, S., Shah, A., Wang, Z., Safikhani, A., & Kamga, C. (2020). Taxi and mobility: Modeling taxi demand using ARMA and linear regression. Procedia Computer Science, 177, 186-195.
HandyDART. Retrieved from https://www.translink.ca/rider-guide/transit-accessibility/handydart
Ismaeil, H., Kholeif, S., & Abdel-Fattah, M. A. (2022). Using Decision Tree Classification Model to Predict Payment Type in NYC Yellow Taxi. International Journal of Advanced Computer Science and Applications, 13(3).
Ke, J., Zheng, H., Yang, H., & Chen, X. M. (2017). Short-term forecasting of passenger demand under on-demand ride services: A spatio-temporal deep learning approach. Transportation Research Part C: Emerging Technologies, 85, 591-608.
Lee, E., Choi, H., & Kim, D.-G. (2023). PGDRT: Prediction Demand Based on Graph Convolutional Network for Regional Demand-Responsive Transport. Journal of Advanced Transportation, 2023.
Li, S., & Zhang, X. (2020). Research on orthopedic auxiliary classification and prediction model based on XGBoost algorithm. Neural Computing and Applications, 32, 1971-1979.
Liao, S., Zhou, L., Di, X., Yuan, B., & Xiong, J. (2018). Large-scale short-term urban taxi demand forecasting using deep learning. Paper presented at the 2018 23rd Asia and south pacific design automation conference (ASP-DAC).
Liyanage, S., Abduljabbar, R., Dia, H., & Tsai, P.-W. (2022). AI-based neural network models for bus passenger demand forecasting using smart card data. Journal of Urban Management, 11(3), 365-380.
Ma, T. M., Yamamori, K., & Thida, A. (2020). A comparative approach to Naïve Bayes classifier and support vector machine for email spam classification. Paper presented at the 2020 IEEE 9th Global Conference on Consumer Electronics (GCCE).
Menninger-Mayeda, H., Berger, P. M., Berger, D. E., McCormick, B., & Boyle, D. K. (2004). Demand forecasting and the Americans with Disabilities Act: Orange county, California, Transportation Authority's access program. Transportation research record, 1884(1), 55-64.
Pranckevičius, T., & Marcinkevičius, V. (2017). Comparison of naive bayes, random forest, decision tree, support vector machines, and logistic regression classifiers for text reviews classification. Baltic Journal of Modern Computing, 5(2), 221.
Qi, Z. (2020). The text classification of theft crime based on TF-IDF and XGBoost model. Paper presented at the 2020 IEEE International conference on artificial intelligence and computer applications (ICAICA).
Rajendran, S., Srinivas, S., & Grimshaw, T. (2021). Predicting demand for air taxi urban aviation services using machine learning algorithms. Journal of Air Transport Management, 92, 102043.
Shu, P., Sun, Y., Zhao, Y., & Xu, G. (2020). Spatial-temporal taxi demand prediction using LSTM-CNN. Paper presented at the 2020 IEEE 16th International Conference on Automation Science and Engineering (CASE).
Singh, A., Thakur, N., & Sharma, A. (2016). A review of supervised machine learning algorithms. Paper presented at the 2016 3rd International Conference on Computing for Sustainable Global Development (INDIACom).
Stadler, T., Sarkar, A., & Dünnweber, J. (2021). Bus demand forecasting for rural areas using XGBoost and random Forest algorithm. Paper presented at the Computer Information Systems and Industrial Management: 20th International Conference, CISIM 2021, Ełk, Poland, September 24–26, 2021, Proceedings 20.
Toqué, F., Khouadjia, M., Come, E., Trepanier, M., & Oukhellou, L. (2017). Short & long term forecasting of multimodal transport passenger flows with machine learning methods. Paper presented at the 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC).
Wang, Z., Huang, S., Wang, J., Sulaj, D., Hao, W., & Kuang, A. (2021). Risk factors affecting crash injury severity for different groups of e-bike riders: A classification tree-based logistic regression model. Journal of safety research, 76, 176-183.
Wheel-Trans. Retrieved from https://www.ttc.ca/wheel-trans
Wu, J., & Hicks, C. (2021). Breast cancer type classification using machine learning. Journal of personalized medicine, 11(2), 61.
Yu, B., & Xu, Z.-b. (2008). A comparative study for content-based dynamic spam classification using four machine learning algorithms. Knowledge-Based Systems, 21(4), 355-362.
Zhao, Z., Chen, W., Wu, X., Chen, P. C., & Liu, J. (2017). LSTM network: a deep learning approach for short‐term traffic forecast. IET Intelligent Transport Systems, 11(2), 68-75.
大愛電視. (2023). 無障礙服務！復康巴士的便與不便.
中華民國醫師公會全國聯合會. (2019). 各鄉鎮各類醫療機構數.
白峻安. (2014). 利用類電磁演算法求解於需求反應式撥召問題－以新北市復康巴士為例.
立法院法制局. (2021). 復康巴士服務相關法制之探討.
交通部運輸研究所. (2021). 預約式通用小客車運輸服務之試辦與推廣應用.
闫贺, 朱丽, & 戚湧. (2019). 基于改进 SVM 的城市道路短时交通状态预测. 南京师大学报: 自然科学版, 42(3), 129-137.
吳姿樺. (2021). 運用注意力機制之雙向長短期記憶法於高速公路尖峰時段交通預測之改善.
张军, 胡震波, & 朱新山. (2017). 基于 AdaBoost 分类器的实时交通事故预测. 计算机应用, 37(1), 284-288.
香港復康會. Retrieved from https://www.rehabsociety.org.hk/zh-hant/
陳志華, 楊子緯, 張訓楨, & 賴永崧. (2016). 特徵分析和機器學習方法應用於肝臟疾病檢測. 福祉科技與服務管理學刊, 4(3), 417-429.
陳品齊. (2017). 復康巴士成本函數與經濟特性分析.
陳柔安. (2019). 復康巴士派遣最佳化決策支援系統_ 以花蓮縣派遣數據分析為例.
陳栢睿. (2015). 台灣鐵路管理局線上訂票系統之模擬與政策分析.
陳淑惠. (2018). 基隆市復康巴士服務之探討.
陳靜怡. (2019). 基於集成學習框架之信用違約預測-以信用卡客戶為例.
黃彩珉. (2020). 以使用頻率為基礎之復康巴士派遣規則 以花蓮縣為例.
楊朝傑. (2009). 探討不同障別之身心障礙者使用復康巴士之頻率及使用目的-以台北市伊甸無障礙交通服務為例.
楊繼勝. (2021). 以時空與外在因素建構捷運進出站人數預測模型：基於混合深度學習方法.
經濟學人智庫. Retrieved from https://www.tica.org.tw/article/detail/330
鄭超一. (2015). 復康巴士旅次需求之空間分析-以新北市為例.
戴靖宇. (2017). 應用開放街圖資料最佳化花蓮縣復康巴士派遣問題.
顏佳瑄. (2018). 復康巴士服務再造.