Akademik Veri Yönetim Sistemi
Applied Soft Computing, cilt.1, ss.1-10, 2026 (Scopus)
This study presents a multi criteria decision making (MCDM)-integrated machine learning approach for accurate and reliable clustering of manufacturing firms according to their eco-innovation capabilities. The approach consists of four main phases: data obtaining and preparation, clustering, calculation of performance indices, and performance evaluation of clustering algorithms via MCDM techniques. The required data were collected from manufacturing companies operating in Türkiye using a questionnaire through e-mail and face-to-face interviews. The questionnaire includes 16 measurement variables under three dimensions of eco-innovation capability: product capability, process capability, and organizational capability. For classification of the companies, six clustering algorithms were utilized; K-Means from partitioning-based methods, FCM from fuzzy-based methods, AGNES, SPECTRAL, and BIRCH from hierarchical-based methods, and Gaussian Mixture from distribution-based methods. Subsequently, eight performance indices were calculated to measure the performance of the clustering algorithms. In the last phase of the methodology, three MCDM methods, WASPAS, VIKOR, and TOPSIS, were exploited to determine the best clustering algorithm in a reliable and robust manner according to eight performance indices. At the end of the study, AGNES was revealed as the best clustering algorithm according to all of the MCDM methods, and manufacturing companies were divided into five classes according to their eco-innovation capabilities. The findings of this study provide empirical support for understanding the eco-innovation capabilities of manufacturing firms in Türkiye and form a basis for comparison with companies in other regions. The results also demonstrate the effectiveness of the proposed MCDM-integrated procedure.