Performance Analysis of Naïve Bayes for News Text Classification

Abstract

This study proposes a multinomial Naïve Bayes Classifier technique for identifying news categories and analyzing their classification. To achieve the objective, 1490 data (BBC News) are employed, including 336 business categories, 261 technology categories, 274 politics categories, 346 sports categories, and 273 entertainment categories. The datasets' performance is evaluated using accuracy, recall, precision, and the F1-score. These data are utilized to train the model, resulting in 98.85% and 97.09% accuracy on train and test data, respectively, with an 80-20 split.