Soft Clustering Technique on Academics Performance Evaluation

Abstract

Clustering techniques are  unsupervised learning methods  of mining complex and multi-dimensional data sets such that observations in the same cluster are similar in some sense.  The  student  academic  performance  evaluation  problem  can  be  considered  as  a clustering problem where clusters are formed on the basis of students intelligence. Choosing the  right  clustering  technique  for  a  given  dataset  is  a  research  challenge.  Therefore,intelligence-based  grouping  is  essential  for  maintaining  the  homogeneity  of  the  group; otherwise it would be difficult to provide good educational recommendation to the highly diverse  student  population.  Homogenous  grouping  of  students  with  similar  result  ranking into   classes  would  further  make  student  academic  performance  analysis  detailed  and sufficient  for  recommendation.  Grouping  of  students  using  Fuzzy  C-Means  (FCM) techniques  with  the  level  of  their  degree  of  membership  into  different  clusters  allows  for overlapping of boundaries and resolve sharp boundary  problems  as opposed to crisp-based method. FCM technique will reveal the degree of membership trend in the clusters which is the focus of this work. In  this work, we implemented Soft clustering technique (Fuzzy CMeans)  in  C++  for  student  academic  performance  analysis.  This  will  proffer recommendations that will enhance student performance.