Ability estimation in computerized adaptive test using Mamdani Fuzzy Inference System

TL;DR: In this article, the Mamdani Fuzzy Inference System (MFIS) is used to estimate the ability of students based on the parameters of the questions and answers given.

Abstract: Assessment is an activity to find out the learning outcomes of a course. Conventionally, assessment is done using a pencil and paper test. Along with the development of information and communication technology, evaluation can be done computerized, known as Computer Based Test (CBT). The Computerized Adaptive Test (CAT) is one form of CBT where the items given are chosen based on the students' abilities. This research aims to design a system that can estimate the ability of students based on the parameters of the questions and answers given. The estimation method uses the Mamdani Fuzzy Inference System (MFIS). The input MFIS is a level of difficulty and discrimination of the questions, the probability of students being able to answer correctly, and the student's answer, while the output is an ability estimated. Based on this fuzzy system output, the next item questions will be determined according to the ability of the students. 24 IF-THEN rules are used for fuzzy systems. CAT simulations are carried out for Linear Algebra Course with six topics, namely vector, matrices concepts, matrices operation, determinant, inverse, and matrices applications. The type of question given is multiple choice. Giving items will be stopped if the value of the estimated ability of the students has not changed. From the simulations carried out, for each topic, the ability of the student can be acquired with the number of questions as many as six questions. So that the CAT system can minimize the time of the exam, reduce the subjectivity of the assessment and can arrange for each student to get the questions according to his abilities.