一、研究目的:本研究以電商網站的商品圖片配置及庫存管理問題為研究議題,參考學者所發展之VADD模式進行模式建構,並應用近年較新穎的離散型教與學演算法進行解題,再透過小型、中型及大型三種規模問題集進行實驗,遂將結果與離散型基因演算法比較與分析演算法間之解題品質及效率。 二、研究方法:為適用本研究問題的離散型決策變數,遂參考學者所提之變數型態轉換技巧,將傳統教與學演算法及基因演算法轉換為離散型教與學演算法及離散型基因演算法,並以實驗進行比較與分析。 三、研究結果:本研究實驗分析結果得悉,離散型教與學演算法除所需設定控制參數較少外,在本研究所進行的實驗中其求解品質和效率較離散型基因演算法為佳,故建議未來業者可將離散型教與學演算法作為主要的問題求解方法。 四、研究貢獻:本研究將傳統教與學演算法及基因演算法轉換成離散型教與學演算法及離散型基因演算法,以適用在離散型決策變數的最佳化模型求解,此外,在實務上離散型教與學演算法可協助電商業者在短時間內決策商品圖片配置及庫存,並提高商品銷售額及企業獲利。
Purpose─This research aims to propose a discrete teaching learning based optimization (DTLBO) method to solve the visual-attention-dependent demand (VADD) model for the product placement and inventory management problem of e-commerce websites. Method─In order to verify the usefulness and efficiency of the proposed method for the VADD model, this research compares the proposed method and the discrete genetic algorithm (DGA) method with large, medium and small problems. Result─Research results show that the efficiency of the proposed method is better than the DGA method in terms of their average execution time and the approximation ratio. Contribution─This research presents a DTLBO method to solve the VADD model that requires fewer control parameters than other optimization algorithms available in the literature. This method can help e-commerce retailers rapidly making decisions on product placement and inventory management to increase the sales of goods and profits of enterprises.