Imbalance Data Handling using Neighborhood Cleaning Rule (NCL) Sampling Method for Precision Student Modeling

Student modeling has an important role in every educational process. In general, educational process begins with student admission process, teaching and learning process, and assessement of the learning outcomes. These sequential processes can be represented as a data or called as the Educational Data (ED). However, in real life, the Educational Data has unbalanced characteristics. To overcome the imbalanced issue, some balancing methods are applied. The balancing process basically divided into three methods: undersampling, oversampling and hybrid of oversampling and undersampling. In this paper, we focus on balancing the Educational Data using the undersampling approach Neighborhood Cleaning Rule (NCL) to obtain the Precision Student Modeling. Data that has been undersampled using NCL is then classified pusing the Decision Tree C4.5 algorithm. While, the performance evaluation is processed pusing the accuracy calculations. The test result using NCL shows an accuracy value of 91.37%. The value of accuracy from the research is represented of the student who fail and succeed academically, so that appropriate treatment can be given. This accounting value obtains the standard error in the educational application (10%).