本研究旨在預測台灣地區280kgf/cm2預拌混凝土的價格變動，並探討其影響因素，以提供業界與政府更精準的價格趨勢分析。本研究運用決策樹CART演算法模型(Classification and Regression Tree, CART)及類神經網路模型(Artificial Neural Network, ANN)兩種機器學習模型，分析2018年至2023年間的相關數據，並預測未來三年(2024年至2026年)之價格變化。研究結果顯示，影響預拌混凝土價格的主要因素包括水泥物價指數、砂石級配指數、預拌混凝土物價指數、營造工程總指數、工業生產總指數、原油均價及匯率七項因素。此外，在模型的比較結果顯示，決策樹CART演算法模型提供較佳的可解釋性，而類神經網路則在預測中表現較優。綜合兩者之優勢，可提升價格預測的準確度。
預測模型分析如下，北區在北區類神經網路四年期模型表現最佳，誤差指標均最低。中區在中區類神經網路六年期模型的表現最佳，R2值達0.9941。南區類神經網路模型在各年期預測表現均優於決策樹CART演算法。花東地區在花東地區決策樹CART演算法二年期模型表現最佳，其次為花東地區類神經網路模型六年期。平均價格的部分在平均價格類神經網路四年期模型表現最佳，R2值達0.9965，MAPE低至0.47%。
本研究針對北區、中區、南區、花東地區及全區平均之預拌混凝土價格，分別建構類神經網路與決策樹CART演算法模型進行2024年至2026年價格預測。結果顯示，各區預測價格普遍低於2024年實際值，模型呈現保守預測傾向。各區MAPE皆低於10%，顯示具高精準度，中區(MAPE 1.08%)與北區(MAPE 1.79%)表現最佳，花東地區與南區之誤差相對略高，仍屬可接受範圍。

Ready-mixed concrete(RMC) with a specified compressive strength of 280 kgf/cm² is a fundamental construction material widely used in infrastructure projects and real estate development. This study focuses exclusively on RMC within the main island of Taiwan, excluding offshore islands and other countries. Price fluctuations of RMC significantly affect project costs, budgeting accuracy, and overall market stability. Various external factors such as raw material price volatility, changes in government policy, and international economic shifts contribute to the uncertainty of RMC pricing. In response to these challenges, this study develops predictive models using Classification and Regression Tree (CART) and Artificial Neural Network (ANN) methodologies to forecast RMC prices in Taiwan for the years 2024 through 2026, based on data collected from 2018 to 2023. The models incorporate key economic indicators including cement price index, aggregate price index, construction price index, industrial production index, crude oil prices, exchange rates, and regional economic conditions. The results demonstrate that ANN models generally yield superior predictive accuracy, whereas CART models offer the advantage of higher interpretability. These findings provide a robust analytical foundation to aid government agencies and industry stakeholders in effective cost management and strategic planning.

1.Al-Araidah, O., Momani, A. M. S., Albashabsheh, N., Mandahawi, N., & Fouad, R. H. Costing of the production and delivery of ready-mix concrete. Jordan Journal of Mechanical and Industrial Engineering, 6(2), 163-173. (2012).
2.Elshakour, H. A., Abdel-Razek, R. H., & Abdel-Samad, D. Predicting production rate of pouring ready-mixed concrete using neural networks. Proceedings of the 6th Alexandria International Conference on Structural and Geotechnical Engineering, Alexandria University, Egypt. (2007).
3.Liu, Q., Huang, M., & Lee, W.-S. Forecast of concrete price movement based on time series and improved random forest model. Operations Research and Management Science, 33(6), 132-138. (2024).
4.Megantara, B., & Melinda, T. The influence of product quality, price, brand image and service on purchase decisions for dry type ready mix concrete: Case study at PT. XYZ East Java. International Journal of Review Management Business and Entrepreneurship (RMBE, 2)(1), 188–196. (2022).
5.Onwuka, D. O., Okere, C. E., & Onwuka, S. U. Prediction of concrete mix cost using modified regression theory. Nigerian Journal of Technology (NIJOTECH, 32)(2), 211-216. (2013).
6.Ranti, H., Yusuf, L., & Rossa, N. H. Raw material cost prediction planning and ready-mix product sales using adaptive linear neural network method. Proceedings of International Conference on Multidisciplinary Research for Sustainable Innovation, 1(1). (2024).
7.Sakhare, V., Taware, T., Ingole, R., Bansode, V., Karade, M., & Kulkarni, Y. Investigating the factors affecting sustainability of ready-mix concrete plants: Case study of Pune region. Discover Civil Engineering, 1(106). (2024).
8.Sun, C., Wang, K., Liu, Q., Wang, P., & Pan, F. Machine-Learning-Based Comprehensive Properties Prediction and Mixture Design Optimization of Ultra-High-Performance Concrete. Sustainability, 15(21), 15338. (2023).
9.行政院主計處網站. (n.d.). Available from https://www.dgbas.gov.tw/
10.中華民國統計資訊網. (n.d.). Available from https://nstatdb.dgbas.gov.tw/dgbasall/webMain.aspx?k=main
11.公共工程雲端服務網. (n.d.). Available from https://pcic.pcc.gov.tw/pwc-web/
12.財政部南區國稅局全球資訊網. (n.d.). Available from https://www.ntbsa.gov.tw/
13.李一平。營建物價指數對六都營建業營業額之影響分析 (碩士論文)。國立高雄科技大學國際企業研究所。(2023)
14.李祖源。運用決策樹和類神經網路預測台灣鋼筋價格 (碩士論文)。國立成功大學土木工程學系研究所。(2023)
15.李詩琪。營建材料價格的調整對房價的影響 (碩士論文)。國立臺北科技大學高階土木營建工程管理雙聯碩士學位學程。(2023)
16.張靖民。原油與金屬價格波動對台灣產業指數之影響 (碩士論文)。國立中興大學財務金融學研究所。(2022)
17.黃智穎。台南市成大城房價之預測 (碩士論文)。國立成功大學土木工程學系研究所。(2021)
18.謝尚瑾。國際原油價格波動與台灣產業指數之關聯 (碩士論文)。國立中正大學財務金融學研究所。(2018)
19.翁志青。營造工程物價指數波動對鋼筋及水泥材料價格之影響 (碩士論文)。國立中興大學土木工程學系研究所。(2017)
20.陳冠廷。以類神經網路預測台南市東區的房價 (碩士論文)。國立成功大學土木工程學系研究所。(2019)
21.蔡景丞。使用類神經網路預測蔬菜的價格與銷售量-以西螺果菜市場及某蔬果批發商為例 (碩士論文)。國立雲林科技大學工業工程與管理系研究所。(2019)
22.黃馨萱。建立複合長短期記憶網路預測離島混凝土物價模型 (碩士論文)。國立金門大學土木與工程管理學系碩士班。(2024)
23.鄭憲忠。臺灣地區預拌混凝土及其材料之價格波動分析 (碩士論文)。中華大學土木與工程資訊學系碩士班。(2010)
24.張啟昇。應用灰色理論於預拌混凝土價格預測之研究 (碩士論文)。佛光大學經濟學系研究所。(2009)