Sentiment Analysis of Acceptance TVET Online Courses on the Skill Academy App from Google Play: Leveraging Text Mining with Comparison Machine Learning Model

<ns3:p>Background The rapid growth of online learning platforms has transformed Technical and Vocational Education and Training, enabling broader access to skill based education through mobile applications. Understanding user acceptance is essential to ensure the sustainability and effectiveness of digital TVET platforms. User reviews available on application marketplaces provide valuable insights into learners’ perceptions, satisfaction, and challenges encountered during use. Methods This study analyzed 3,000 user reviews of the Skill Academy application collected from the Google Play Store. Text preprocessing included data cleaning, case folding, tokenization, filtering, stemming, and translation into English to ensure compatibility with the Valence Aware Dictionary and Sentiment Reasoner. Sentiment labels were generated using VADER and categorized into positive, neutral, and negative classes. Text features were extracted using Term Frequency Inverse Document Frequency. Six machine learning classifiers were evaluated, namely Naive Bayes, Support Vector Machine, Logistic Regression, Random Forest, Decision Tree, and K Nearest Neighbors, using a 70:30 train test split. Results The sentiment analysis results showed that 69.4 percent of the reviews expressed positive sentiment, 20.5 percent were neutral, and 10.1 percent were negative. Positive reviews predominantly emphasized course usefulness, ease of use, and skill development benefits. Negative reviews were mainly associated with technical issues such as application errors and performance limitations. Among the evaluated models, the Support Vector Machine achieved the best performance, with an accuracy of 85.77 percent and an area under the curve value of 0.97. Conclusions The findings indicate a high level of user acceptance of online TVET courses offered through the Skill Academy application, primarily driven by content quality and usability. Nevertheless, addressing technical performance issues remains essential to improve user satisfaction and support sustainable long term adoption. The results also demonstrate that the Support Vector Machine model is highly effective for sentiment classification in the context of online TVET platforms.</ns3:p>