Analisis Sentimen Dalam Pengkategorian Komentar Youtube Terhadap Layanan Akademik dan Non-Akademik Universitas Terbuka Untuk Prediksi Kepuasan

Abstract

The key to the success of an educational organization in achieving its goals of course cannot be separated from the quality of service both in academic and non-academic forms. Where in achieving these goals of course by giving satisfaction to the academics. The case study was carried out to predict service satisfaction at the Open University by using comments on social media Youtube as data processing. The text mining approach is a good alternative in terms of interpreting the meaning in the comments given. This study aims to analyze the predictions of service satisfaction from several categories as a benchmark. The categories are: Module, Tutorial, Scholarship, Lecturer, Exam, Application, Non-Academic and Others. The research method used is comparative, by applying 4 algorithms, namely Decision Tree (DT), Support Vector Machine (SVM), Naïve Bayes (NB) and Random Forest (RF) for Prediction Accuracy. The total initial dataset is 7776 data and after cleansing and preprocessing is 6920 data. And then evaluated for the 7 categories after being accured to produce: Module category with the highest accuracy of 99.37% using the DT algorithm, Application Category with the highest accuracy of 100% using the DT algorithm, Teaching Category the highest accuracy of 99.42% using the algorithm DT. The tutorial category has the highest accuracy 92.4% using the SVM algorithm, the exam category has the highest accuracy 99.7% using the RF algorithm, the non-academic category has the highest accuracy 99.90% using the DT algorithm. And for the Others category the highest accuracy is 96.58% using the DT algorithm