The Association Rule Method for Mapping and Recommendation System on Students’ Difficulties

Abstract

The evaluation process becomes a very important factor because with this stage information about the quality of education can be known, so that the education system can be improved continuously. However, data resulted by the test as a part of the educational evaluation have not been comprehensively analyzed. Therefore, this research is aimed to design and implement a model based on Machine Learning (i.e., the apriori algorithm) that can be used for mapping students' difficulties and making recommendations. In learning, we consider the following steps. Firstly, some multiple-choice items/questions are collected, and then each of them is mapped to syllabus/curriculum. After that, students are tested by these items, so that we obtain some data containing a list of ID items with wrong answers by each student. Then, by using the algorithm, we build a set of rules representing association of the ID items in data training. Along with relations between each item and its syllabus, the rules become the model in the learning phase. After obtaining the model, we predict difficulties of some students by matching their wrong answers with the rules, and then take a look at the relations between ID items in fired rules with syllabus. So, we can analyze students' difficulties according to the frequent items, and then make some recommendations. Moreover, we perform some experiments on 33 students in a vocational high school in Bandung, Indonesia. Results of the systems are validated by human experts. Finally, we can state that the objectives have been accomplished successfully.