EVALUATION OF TEXT CLASSIFICATION ALGORITHMS: A STUDY ON CLASS IMBALANCE AND DATA PREPROCESSING

Abstract

This paper presents a comparative analysis of several Machine Learning algorithms in the thematic classification of comments across various media. Using a dataset that shows a significant class imbalance problem, the study examines the effectiveness of each model under different preprocessing conditions. The results show that the Random Forest algorithm is the most suitable, as it can handle correlated features with an accuracy of 0.92, while Naive Bayes faces limitations with data that has low representativeness or high ambiguity. This research highlights the importance of preprocessing in text analysis and provides insight into the performance of different algorithms in text classification.