SOFT SKILLS SEGMENTATION AND PROFILING ANALYSIS

Abstract

Internship is an intervention program to improve student' soft skills. Soft skill is crucial in preparing them for the workforce and helping them gain a better understanding of the workplace. Eventually the results of how well undergraduate students have used their soft skills in a real-world work environment is reported, but which students with low score soft skills are not being identified. By having that information, higher institution can help in preparing special training for certain soft skills with low score after they have finished their internship. With clustering techniques, segmentation, and the division of a large group of students into more focused subgroups (or cluster) based on shared soft skills can be accomplished. In this study, a segmentation of the internship students according to their soft skills is investigated. In this study, before implementing the outperform algorithm into real data, 20 hierarchical clustering algorithms were compared using cophenetic correlation coefficient. The 81 real data based on 12 soft skills were analysed using MATLAB software with the combination of Euclidean, Squared Euclidean, Spearman, Correlation and Minkowski distances with four linkages methods which are single, complete, average and Ward’s method. The results show that the combination of Euclidean and Minkowski distance techniques with average-linkage method are the best combination with the cophenetic correlation coefficient value of 0.8412. Overall, it shows that S1 (knowledge and understanding skill) had the highest score among the three clusters produced with the value of 9.650 while S4 (communication skill) has the lowest score between other skills with the value of 4.333 respectively. As a summary, it has shown that clustering approaches can be used to explore students' soft skills and understanding their profiling.