Partitioning the stocks of a portfolio by k-medoids clusteringapproach

Machine learning is mainly used in practice because of the existence of large set of data.The target of this article is to study partitioning a large set of stocks inside a portfolio bythe simple yet efficient k-medoids procedure. An algorithm is developed for this purpose.The unsupervised model is capable to receive financial returns and to illustrate the most andleast risky clusters of stocks to manage the risk.