A Comparative Study of Logistic Regression, Support Vector Machines, and LSTM Networks for Sentiment Classification in Academic Reviews

In recent years, sentiment analysis has become an important tool for understanding what people think and feel about different topics, especially in education. This study looks at how well Long Short Term Memory (LSTM) networks work compared to traditional methods like Support Vector Machines (SVM) and Logistic Regression for classifying sentiment in text. By analyzing a dataset of 18,901 college reviews, which includes reviews sourced from Kaggle and additional reviews for improved dataset balance, this research evaluates how well LSTM networks can detect nuanced sentiments compared to more conventional models. The study includes steps like cleaning the data, testing and fine-tuning several machine learning methods, and measuring how accurately they capture the contextual nuances in the reviews. Findings reveal that LSTM networks outperform traditional methods, offering enhanced accuracy in capturing the emotional depth and context of the text. This study provides valuable insights into the use of advanced neural network models for sentiment analysis, highlighting their ability to improve the interpretation of user feedback. The findings suggest that incorporating LSTM networks into educational feedback systems could result in more accurate sentiment classification and contribute to improved decision-making processes.