<p>Supplier selection became complex due to increasing in the data size of criteria and alternatives, and the present studies failed in the integration of multi-criteria decision-making (MCDM) with advanced clustering (unsupervised machine learning). To fill this gap, the author proposed a novel clustering-based three-phase C-MCDM structure for enhancing the supply chain accuracy and reducing the complexity of the supplier selection problem. In phase I, the criteria are defined and selected; in phase II, the suppliers are divided into groups with the help of selected clustering algorithms, and the best cluster is identified. Finally, in the third phase, multi-criteria classification of suppliers is performed on the best cluster to obtain the ranking of suppliers to proceed further. The author implemented C-MCDM approach in a case study of the textile industry with 362 suppliers, and they are selected based on social, economic and environmental criteria along with the technical and supplier’s profile criteria. These criteria are further subdivided into 34 criteria based on literature, customers and expert managers. The rank of suppliers is obtained by Entropy and DEMATEL method with the integration of DAHP, TOPSIS and VIKOR methods. It is observed that application of clustering in conjugation with the usual MCDM methods is an attractive option for decision-making. In the end of study an ensemble ranking method named HQ-R, based on the half-quadratic (HQ) theory is employed to estimate the final ensemble ranking, achieving a consensus index of 0.9822 and a trust level of 0.9895, calculated to determine the agreement index and reliability in the ensemble results.</p>