Implementasi Hadoop Mapreduce Untuk Memprediksi Predikat Kelulusan Mahasiswa

Muhammad Awaluddin (Universitas Muhammadiyah Buton)
Rini Angelia Mahlil (Universitas Muhammadiyah Buton)
Darmawan Darmawan (Universitas Muhammadiyah Buton)
La Ode Muhammad Saidi (Universitas Muhammadiyah Buton)

Journal on Education
Vol 5 No 4 (2023): Journal on Education: Volume 5 Nomor 4 Mei-Agustus 2023

This study aims to implement Hadoop MapReduce as a data processing framework in order to predict student graduation predicates. The growth in the number of students in higher education institutions has led to an increasingly large and diverse volume of academic data. The use of Hadoop MapReduce is expected to overcome the challenges of large-scale data processing and support the development of an efficient pass rate prediction system. This research method involves analyzing student academic datasets, which include information such as exam results, course grades, classes taken, and other academic attributes. This data represents the population of students enrolled in a particular study program at higher education institutions. Hadoop MapReduce is implemented to process data in parallel, using the Map and Reduce functions, as well as setting the appropriate configuration so that the system can handle large-scale data processing efficiently. This study aims to implement Hadoop MapReduce as a data processing framework in order to predict student graduation rates. The use of Hadoop MapReduce is expected to overcome the challenges of large-scale data processing and support the development of an efficient pass rate prediction system. The MapReduce Framework execution process was carried out very quickly, with the longest execution time of 2,375 milliseconds and the fastest 105 milliseconds, for processing student GPA data and the number of student credits, respectively. The prediction test results for 2015 graduates showed a difference of 23 students between the prediction results and reality, with a prediction error rate of 2.09%. Hadoop MapReduce is able to handle data processing very efficiently, even for relatively small data. In addition, the prediction of student graduation rates using the implemented system gives adequate results with a low error rate. Nevertheless, it is necessary to carry out further evaluation and improvement of the prediction model to improve prediction accuracy. Thus, Hadoop MapReduce remains the right choice for big data analysis and processing in various applications, including in the context of predicting student graduation rates.