Analisis Sentimen Penghapusan Skripsi sebagai Tugas Akhir Mahasiswa Menggunakan Metode Multi-Layer Perceptron

Abstract

Indonesia has levels of education where one of them is undergraduate education. There are requirements that must be done to get a bachelor's degree, one of which is to complete a final project in the form of a thesis. Nadiem Makarim, Minister of Education, Technology and Higher Education in his speech announced a new policy in the field of education regarding the non-obligation for students to prepare a thesis as a requirement for graduation. Based on this, there are pros and cons from the community, the sentiment analysis process related to this is needed. This research aims to map public sentiment contained in TikTok and YouTube social media related to the elimination of thesis using the MLP method. The stages carried out consist of observation, data collection, labeling, data normalization, preprocessing, data partitioning, TF-IDF weighting, classification, and evaluation. The accuracy obtained at the preprocessing scenario stage is 86% with case folding and stemming scenarios. Furthermore, this scenario is used in testing based on data partitioning where the highest accuracy results are obtained with a portion of 90% training data and 10% test data. The accuracy obtained is 94%.