Fuzzy K- means cluster validation for institutional quality assessment

The most important facts in educational institutional system growth lies in the quality of services rendered. (i.e., faculty profile, student performance and infrastructure requirements). The highest level of quality in educational institution can be achieved by utilizing the managerial decision makers with valuable implicit knowledge, which is currently unknown /hidden to them. The knowledge hidden among the educational data set is extractable through data mining technology. Clustering, an unsupervised learning depends on certain initiation values to define the subgroups present in a data set. Based on the features of the dataset and input parameters cluster formation may vary, which motivates the clarification of cluster validity. The proposed work presented a fuzzy k-means cluster algorithm in the formation of student, faculty and infrastructural clusters based on the performance, skill set and facilitation availability respectively. With the obtained data clusters, quality assessment is made by predictive mining using decision tree model. The cluster validation criterion is introduced to find the optimal input metrics for fuzzy k-means algorithm. Validation criteria focus on the quality metrics of the institution features for cluster formation and handle efficiently the arbitrary shaped clusters. Experimental results show improved stability and accuracy for clustering structures obtained via sub sampling, and adaptive techniques. These improvements offer insights into specific decision within the data sets. The experimental results confirm the reliability of validity index showing that it performs favorably in all cases selecting independently of clustering algorithm the scheme that best fits the data under consideration.