Feature Selection on Grouping Students Into Lab Specializations for the Final Project Using Fuzzy C-Means

Abstract

The student’s Final Project is critical as a requirement to graduate from the University. In the PSTI at Mataram University, each student is required to choose a specialization lab to focus on the final project topic that they will work on. From the questionnaire, 57.7% of students answered that it is difficult to select a lab, and others answered that they prefer to determine the labs based on the grades of the courses that represent each lab. This research aimed to group and analyze students in the final project specialization lab by using the main method, namely Fuzzy C-Means (FCM). The methods used were FCM for clustering, Silhouette Coefficient for analysis of cluster quality results, Pearson Correlation, and Principal Component Analysis for the feature selection processing. The results of this study showed that the FCM method followed by a method for feature selection has better results than previous studies that used the K-Means method without feature selection; with this research result using 131 data, the cluster validation result is 0.501, after feature selection using Pearson correlation is 0.534. Thus, Fuzzy C-Means followed by the right feature selection method can group students into specialization laboratories with good results and can be further developed.