As a result of the use and access of social media, it also has an impact on increasing the amount of data and information, especially text data. Text has become one of the most natural forms of data that is stored, so that the field of text mining is believed to be an advanced field of data mining. Facts that emerge from research studies that have been conducted show that 80% of company information is presented in text documents. Text mining is a multidisciplinary field, involving information retrieval, text analysis, information extraction, and clustering. The text mining classification method is one technique that can be used to carry out classification. Text classification specifically works to group text documents based on categories, and within the scope of news datasets, categories are generally divided into politics, economics, military, sports and others. Statistical methods are one of the most frequently applied methods in text emotion classification. As a method in statistics, Naive Bayes is a classification algorithm that is easy to understand in text classification. Apart from that, Naïve Bayes has good classification effects and performance for processing large-scale data. The conclusion of this research is, Naïve Bayes gets an accuracy value of 77.78%. Random Forest gets an accuracy of 70.1%. KNN gets an accuracy of 24.88% and SVM gets an accuracy value of 80.60%. Meanwhile, the respective running times are Naïve Bayes 0.046 seconds, Random Forest 150 seconds, KNN 15 seconds, and SVM 0.43 seconds.