Evaluating the Performance of Non-banking Financial Institutions by the Means of C-Means Algorithm

We use C-Means clustering algorithm to assess comparatively the performance of non-banking financial institutions (NFIs) in Romania. Firstly, we consider this real world application as a knowledge discovery problem (Data Mining) and we engage in a thorough literature review regarding the application of Data Mining methods in assessing comparatively companies’ financial performance. Then, we apply C-Means on our NFIs’ performance dataset. Next, we compare our results with those obtained when we apply a neural network-based clustering algorithm, called Self-organising Map (SOM) algorithm. The results show that C-Means can offer a good alternative to more sophisticated algorithms in terms of accuracy of the obtained clusters. However, the visualization capability of SOM should be taken into account when we are interested in how companies evolve over time.