A Hybrid Clustering Model for Determining Financially Needy Students for Sponsorship

Abstract

Many organizations in the world are seeking effective means to provide financial assistance to brilliant but needy students. Many research works have been conducted to determine needy students based on classification methods which fail to separate financially stable students from needy ones as the class labels for the classification methods are affected by manual reviews, randomness and the discretion of the selection committee. This paper presents a hybrid clustering method for determining needy students based on income and expenditure using the K-Means and Expectation-Maximization methods with the logical AND operator to eliminate the limitations associated with the classification methods. The study reveals that hybridizing clustering methods yield better results. Also, for a normalized data categorization, the Euclidean distance gives better results than the Manhattan distance in the K-Means algorithm.