Abstract - In this study, we focus on sentiment analysis, an essential technique in the rapidly evolving field of text analytics.
Ourapproach involves preprocessing the movie review text data using tokenization, lemmatization techniques, and feature
extraction using Word of Bags and TF-IDF. We employ three popular machine learning methods, Logistic Regression, SVM,
and Random Forest, to develop sentiment classification models. Our results show that logistic regression with the TF-IDF
technique and default parameters outperforms the other models in terms of minimizing false positives, with an accuracy of
89.20%, a precision of 88.80%, recall of 89.80%, and an area under the receiver operating characteristics curve (AUC) of
89%. These findings have important implications for improving sentiment analysis and developing more accurate and effective
text analytics tools, contributing to the novelty of the work in the journal fields.