本研究旨在探討能源產業於高度競爭市場環境中的生產安排問題。A公司因採取訂單導向的排程策略，致使生產毛利呈現顯著波動。如何透過統計分析與數學規劃方法提升產品組合決策之毛利表現。研究首先採用分群分析，以掌握不同特性能源產品之生產異質性；其次運用多元線性迴歸分析建立產品組合與毛利表現之預測模型，檢驗各群組在不同產能配置下的貢獻；最後結合線性規劃，構建以毛利最大化為目標之決策模型，並以A公司能源產品廠為案例研究。 研究結果顯示，分群分析與迴歸模型的整合能有效提升產品組合配置的合理性，並在多重產能限制下達成毛利最佳化。分群方法降低了因產品異質性所造成的決策複雜度；多元線性迴歸模型則能精準捕捉各群組對毛利的影響；而線性規劃則在公司有限資源下，提供系統化的最佳化決策支援工具，進而導出最優解。 本研究的主要貢獻包括：第一，提出結合統計分析與最佳化方法之混合模型，以提升能源產業在高度競爭環境下的決策效率；第二，透過案例研究驗證模型之可行性與實務價值，並提供產能限制下具體的操作指引；第三，展現數據分析與數學規劃方法於管理決策支援上的跨領域整合應用。此研究不僅對能源產業具參考意義，亦可延伸至其他製造業，作為產品組合與毛利最佳化之決策依據。;This study aims to examine production scheduling issues in the energy industry under highly competitive market conditions. At Company A， reliance on order-driven scheduling has resulted in pronounced fluctuations in gross profit. To address this challenge， the study investigates how statistical analysis and mathematical programming methods can enhance product portfolio decision-making and improve profit performance. Specifically， cluster analysis is employed to capture the heterogeneity of energy products with differing characteristics; multiple linear regression is then applied to construct a predictive model linking product portfolio configurations to gross profit outcomes， thereby assessing the contribution of each product group under varying capacity allocations. Finally， linear programming is integrated to develop a decision model that maximizes gross profit， with an empirical case study conducted at Company A’s energy product plant. The results indicate that the integration of cluster analysis and regression models can effectively enhance the rationality of product portfolio configuration and achieve gross profit optimization under multiple capacity constraints. Cluster analysis reduces the decision-making complexity caused by product heterogeneity; multiple linear regression models accurately capture the impact of each cluster on gross profit; and linear programming， under the company’s limited resources， serves as a systematic optimization decision-support tool that ultimately derives the optimal solution. The main contributions of this study are as follows: First， it proposes a hybrid model that integrates statistical analysis and optimization methods to enhance decision-making efficiency in the energy industry under highly competitive environments. Second， through a case study， the feasibility and practical value of the model are validated， providing concrete operational guidelines under capacity constraints. Third， it demonstrates the cross-disciplinary integration of data analysis and mathematical programming methods in supporting managerial decision-making. This research not only offers valuable insights for the energy industry but can also be extended to other manufacturing sectors as a decision-making reference for product portfolio and gross profit optimization.
