Indrina Sari, Rika (2025) Analisis Sentimen Terhadap Program Mbkm Ferienjob Pada Twitter/x Dengan Menggunakan Metode NaÏve Bayes Dan Decision Tree. Other thesis, Universitas Islam Riau.

Text 203510482.pdf - Submitted Version Restricted to Registered users only Download (3MB) | Request a copy

Abstract

This study aims to analyze sentiment towards the MBKM Ferienjob program on the Twitter/X platform using three classification methods: Naïve Bayes, Decision Tree, and Convolutional Neural Network (CNN). The analysis involves comparing the performance of these methods in terms of accuracy, precision, recall, and F1-Score to determine the most effective method for user sentiment classification. The dataset used consists of tweets labeled as positive, neutral, and negative. For k=5 validation, the RNN method with 50 epochs achieved the best performance with an accuracy of 77.15%, precision of 79.20%, recall of 77.15%, and an F1-Score of 77.33%. Meanwhile, the LSTM method with 50 epochs ranked second, achieving an accuracy of 69.70%, precision of 73.23%, recall of 69.70%, and an F1-Score of 69.85%. On the other hand, CNN with 10 epochs showed the lowest performance, with an accuracy of 60.35%, precision of 63.35%, recall of 60.35%, and an F1-Score of 59.62%. In k=10 validation, the Decision Tree and RNN methods provided competitive results. RNN with 50 epochs showed solid performance with an accuracy of 73.62%, precision of 77.03%, recall of 73.62%, and an F1-Score of 73.58%. However, the LSTM method with 50 epochs exhibited a significant performance improvement, reaching an accuracy of 76.94%, precision of 79.57%, recall of 76.94%, and an F1-Score of 77.14%, making it the best-performing model. Moreover, the analysis also revealed that k=10 provides more stable and accurate results compared to k=5. This research concludes that the selection of the appropriate model and number of epochs significantly affects sentiment analysis results, with the RNN and LSTM methods showing superior performance in this multi-class classification task.

Actions (login required)

View Item