Implementation of an automated grading system with a subtractive clustering method
Nisa Hanum Harani (a*), Putri Suciani Hafid (b), Cahyo Prianto(a)

a) Politeknik Pos Indonesia
Jalan Sariasih No.54, Sarijadi, Sukasari, Kota Bandung, Jawa Barat 40151
*nisahanum[at]poltekpos.ac.id
b)Politeknik Pos Indonesia
Jalan Sariasih No.54, Sarijadi, Sukasari, Kota Bandung, Jawa Barat 40151
putrisuciani03[at]gmail.com

Abstract

The Trainee participants exam process has been conducted conventionally during the examination process, the scoring/test scores, and the division of high and low scores of participants. The Exam tends to be risky because it spends a long time and high accuracy. In this study built an online exam system that implements auto grading that facilitates the Learning Center and the participants Management Trainee. Regarding the exam process, giving and receiving scores/grades quickly. This research using fuzzy subtractive clustering method, because of fuzzy subtractive clustering able to generate rules without initializing the number of clusters at the beginning of the process. The formation of the number of rules/clusters is influenced by the radius and reject, ratio, while the accept ratio does not affect the cluster results. Fuzzy subtractive clustering is used to define high and low scores of participants. Following the calculation to determine the group value of Management Trainee participants that is by using two parameters of the data used is the value of Written Comprehensive Examination and Oral Comprehensive Examination. The test results showed the cluster that formed as much as 2 groups by using samples of 10 data with the value of each, the test using the value of accept ratio 0.6, rejects ratio 0.2, radius 0.8, and squash factor 1.25.

Keywords: Auto grading, CBT, Fuzzy subtractive clustering, Data Mining

Topic: Computer Science Education