A hybrid fuzzy framework for clustering and ranking listed companies via the K-means algorithm

The extant literature indicates that companies, including those listed on the Tehran Stock Exchange, are typically classified based on their respective industries. The findings of this study indicate that a reclassification of these companies may be warranted if alternative indicators are to be considered in the clustering process. To this end, the present study employs six indicators for the purpose of clustering the companies: price-to-return ratio (P/E), volume, liquidity rating, return on equity (ROE), β coefficient, and return on capital. The values of these indices are extracted for each company, and then the companies are categorized into eight clusters using the K-means method. The significance of these groups varies from user to user, contingent upon the relative importance of the indicators from the user’s perspective. To categorize these clusters, we initially employ the Fuzzy Analytical Hierarchy Process (FAHP) to ascertain the weights of the indexes. Subsequently, we utilize the TOPSIS method to prioritize the clusters. The results indicated that, according to the experts’ assessments, the P/E ratio, beta coefficient, and liquidity rating, with weights of 0.365, 0.221, and 0.212, respectively, are the most significant indicators for the evaluation of the clusters. It is imperative to acknowledge that the decision matrix employed in this study is predicated on the assumption that the centroid of each cluster serves as its representative. The approach delineated in this study has the potential to serve as a foundational element in subsequent research endeavors focused on portfolio construction based on firm similarity.