A Layered KNN-SVM Approach to Predict Missing Values of Functional Requirements in Product Customization

The conversion from functional requirements (FRs) to design parameters is the foundation of product customization. However, original customer needs usually result in incomplete FRs, limited by customers’ incomprehension on these products. As FRs may undermine activities afterwards, managers need develop an effective approach predict missing values FR. This study proposes integrative obtain complete k nearest neighbor (KNN) algorithm employed continuous variables FR, using improved distance formula for two FRs. Support vector machine (SVM) classifiers are adopted classify categorical combined with directed acyclic graph multi-class classification. KNN and SVM then integrated into a multi-layer framework where both exist. A case elevator customization conducted verify that KNN-SVM feasible accurate prediction FR values. Furthermore, outperforms other five single composite methods, average reduction root mean squared error (RMSE) 39% 21% against KNN-Tree, respectively.