Fountain of Informatics Journal

5-9

Authors: Indah Puji Astuti

The student is one of unsure in University or Higher Education. The student has different of data, such as self-identity information such as address, type of school, work of parents, type of class, etc. The students whose graduation rate is different, on time and not on time. The number of students who graduate is not on time will be a problem not only for university but also for faculty. The number of students graduating each year is one of factor of assessment when faculty or study program submits accreditation. C4.5 Algorithm is one of classification algorithm with decision trees. In this study conducted an analysis of student data Engineering Studies Program University of Muhammadiyah Ponorogo 2012/2013. The decision trees in this case is useful for exploring student data, finding the hidden relationship between of input attributes with a target attribute. The input attribute consist of, the type of school, address, work of parent, and type of class. The output attribute to classify is status, which consists of "on time" and "not on time". The results from this analysis shown that in this case the C4.5 algorithm can predict with an accuracy value only 82%.

20-27

Authors: Gita Indah Marthasari

One indicator of college efficiency is the study period of the students. It is important for university managers to improve the ratio of students who graduate on time. This research aims to analyze the characteristics that affect the study period of students from academic data. The methods used are association rule mining (ARM) and clustering. We propose a framework to analyze academic data using ARM and clustering method. ARM method is a method to find association rules that meet minimum support and minimum confidence. The algorithm used is Apriori. While clustering using Simple Expectation-Maximization (EM-clustering) algorithm. Simple EM is a model-based algorithm that searches for maximum likelihood estimation in the probability model. The variables analyzed were student achievement index, province of the students, and type of high school. The analysis is done using WEKA. Research begins with the collection of data from the primary source of the Biro Administrasi Akademik (BAA) Universitas Muhammadiyah Malang (UMM). Then, we do the data cleaning and transformation. Analyzing process is done in two step. First, do a rule search using Apriori algorithm. The regulated parameters are the minimum support and minimum confidence value. Second, we use Simple EM algorithm for the clustering process. The experiments were conducted to find the clustering result with the largest log likelihood value. Based on the experiment, the method used successfully describes the characteristics based on the study period.