A New Fine-Kinney Method Based on Clustering Approach

Abstract

In this study, a new approach to Fine-Kinney risk assessment method is developed in order to overcome the limitations of the conventional method with clustering algorithms. New risk level of classes are attempted to determine with K-Means and Hierarchical clustering algorithms with using two different distance functions which are Euclidean and Manhattan distances. According to the results, K-Means algorithms have provided accurate and sensitive cluster of classes. Classes from conventional and K-Means algorithms are applied and compared to the identified risks of a workshop of a medium sized textile company. Results of the study indicate that clustering techniques are new, original and applicable way to define new classes in order to prioritize risks by overcoming the drawbacks of conventional Fine-Kinney method.