Analysis of Student Sentiment Towards The Digital Literacy Curriculum at University of Singaperbangsa Karawang Using Naïve Bayes

Abstract

Implementation of the Digital Literacy Curriculum at University of Singaperbangsa Karawang (Unsika) as a concrete step considering the importance of digital literacy. Curriculum evaluation is an important step in ensuring the effectiveness of learning. Currently there is no mechanism from Unsika to encourage students to provide feedback on curriculum implementation. One way is by sentiment analysis, which requires sentiment analysis using the Naïve Bayes algorithm which contributes to providing feedback on curriculum evaluation. This sentiment analysis process uses the CRISP-DM (Cross Industry Standard Process for Data Mining) methodology. In an effort to obtain student perceptions, a survey was conducted using a questionnaire. The minimum number of respondents determined using the Slovin formula is 388. In this study the amount of data used was 591. There is an imbalance in the data, to overcome this an oversampling technique can be used using ADASYN (Adaptive Synthetic Sampling Approach). The data has been cleaned to produce 347 positive sentiments and 176 negative sentiments. The results of this research show the best model and word frequency in the form of visualization of words that play an important role in each sentiment category for use in curriculum evaluation. Of the eight model scenarios tested, the model trained with the Naïve Bayes algorithm using a division of 90% training data and 10% testing data with the application of ADASYN became the best model with an accuracy of 89%, precision 100%, recall 85%, and f1-score 92%.