Prediction Model Students’ Performance in Online Discussion ForumFebrianti Widyahastuti PhD Student in Information Technology Deakin University Burwood, Australia +61432509113 fwidyaha@deakin.edu.au Yasir Riady Teacher in English for Librarians Open University Indonesia +6281808332512 yasir@ecampus.ut.ac.idWanlei Zhou Professor in Information Technology Deakin University Burwood, Australia +61392517603 Wanlei.zhou@deakin.edu.au ABSTRACT Prediction is one of the most popular topic in Education Data Mining. Prediction is related to students’ performance, which can be accurately observed through Educational Data Mining. Online discussion forum can utilize prediction modeling techniques to identify key predictors of students’ academic performance by monitoring the progress of students with fully use of online discussion forum. In this paper, a new model in E-learning to predict students’ performance in online discussion forum was proposed. This research was participated to 165 students in English for Librarians course by extracting and analyzing online discussion forum data from E-learning log in Open University, Indonesia. The result from analysis is used to determine which features in online discussion forum as the key indicator to predict students’ performance. Course module instance list viewed, discussion created, discussion subscription created, discussion subscription deleted, discussion viewed, module course viewed, post created, post deleted, post updated, some content has been posted and user report viewed were the features to predicting students’ performance. CCS Concepts: Applied computing→E-learning Keywords Prediction; Discussion forum; Performance; Course Assessment; E-learning; Educational Data Mining; Data Mining. 1. INTRODUCTION There are some increasing research interests in data mining applied in educational sector, it is called educational data mining [1] . Educational Data Mining (EDM) refers to the field of statistics apply, machine-learning, and data mining for various educational data [1]. One of the Educational Data Mining approaches is prediction model. This is a common approach attempting to predict and understand students’ educational outcomes (cf. Romero et al, 2008 ). Prediction models are more demanded for detecting students’ behavior of modelling and assessment functionalities by using those features as models for prediction [2] . There are some models that used for several types of data mining approach to predict the student performance for either current or future outcomes as in Pradeep et al (2015) [3] and Ahmad, F., et al. (2015)[4], they predicted and categorized student performance into three classes: poor, average and good by using comparative analysis of three classification techniques; Rule Based (RB), Decision Tree (DT), Naïve Bayes (NB). There has been limited research using educational data mining for online discussion forum. It is considered new and fresh for researchers to use education data mining associated with communication tools especially online communication in higher education. In the last few years, not many researchers thought and discovered the advantage of the available communication tools related educational data mining. In fact, millions of students from all education levels participate in online communications. Current research do not identify the course assessment reflect with students’ grades. Course assessment is the main part of students’ performance, without course assessment is hard to measure the students’ performance accurately. While many research did not see the gaps between students’ performance and students’ behavior in online discussion forum towards course assessment. Students’ performance is mainly to determine the students’ progress result in academic life, which impacts the status of the students whether to stay or leave the study [5]. This critical issues face by many educational institution, why the predication model is significant important. This is the reason of performance predication that highly important to be implemented in many fields. It is to avoid the reduction risk of students in educational institutions also to improve the quality and students’ skills through helping them to identify the weakness of program study especially for highly risked students. Accordance with the focus of the study, Pearson Correlation was used to find the best predictor in students’ behavior in online discussion forum toward students’ performance. The predicator results can predict students’ performance use the linear regression. The correlation results show that discussion created, discussion subscription created, module course viewed and some content has been posted have strong relationship with assignment 1, assignment 2 and assignment 3. The linear regression as Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions@acm.org. ICIET '17, January 10-12, 2017, Tokyo, Japan © 2017 ACM. ISBN 978-1-4503-4803-4/17/01…$15.00 DOI: http://dx.doi.org/10.1145/3029387.3029393 6
