An improved machine learning-based QFD method for product design using rough set theory and PROMETHEE method

Purpose The main purpose of this study is to develop an improved machine learning-based quality function deployment (QFD) method for identifying the customer requirements (CRs) and measuring the prioritization of the engineering characteristics (ECs) in the product design optimization. Design/methodology/approach The identification of the CRs and the prioritization of the ECs have been explored by integrating the Latent Dirichlet allocation (LDA) model, the rough numbers, the ant colony optimization (ACO) algorithm, and the preference ranking organization method for enrichment evaluations (PROMETHEE) method. Based on the CRs and ECs determined by the LDA model, the rough numbers are introduced to evaluate the CR-EC relationship matrix and the EC-EC correlation matrix. Afterwards, the ACO algorithm is used to optimize the CR-EC relationship matrix, and the PROMETHEE method is applied to measure the prioritization of the ECs. The effectiveness of the proposed method is demonstrated by measuring the prioritization of the ECs in a case study of smartphone design. Findings The results indicate that the improved machine learning-based QFD method significantly improves the prioritization accuracy relative to the existing machine learning-based QFD methods. Originality/value Existing machine learning-based QFD methods suffer from a problem in the representation of uncertain information solely through interval grey numbers bounded by upper and lower limits, which leads to an inaccurate prioritization of the ECs. To address the problem, an improved machine learning-based QFD method is proposed. The main contribution of the proposed method is the integration of the LDA model, the rough numbers, the ACO algorithm, and the PROMETHEE method into the machine learning-based QFD framework. This achieves a more accurate measurement of the importance and the prioritization of the ECs, which improves customer satisfaction and maintains market competitiveness when designing a new product.