Sentiment Analysis Techniques: A Comparative Study of Logistic Regression, Random Forest, and Naive Bayes on General English and Nigerian Texts

Abstract

This research investigates sentiment analysis on two distinct datasets: a general English dataset and a Nigerian dataset (Gangs of Lagos movie review), using three machine learning algorithms: Logistic Regression, Random Forest, and Naive Bayes with python programming language and its libraries. The study aims to evaluate and compare the performance of these models across different linguistic and cultural contexts. Results indicate that Logistic Regression consistently outperforms the other models, achieving the highest accuracy and balanced performance across sentiment classes. Random Forest provides comparable results but struggles with positive sentiment detection in the Nigerian dataset. Naive Bayes shows the lowest overall accuracy, with significant challenges in recall for certain sentiment classes. These findings highlight the importance of model selection and tuning tailored to specific datasets for effective sentiment analysis.