Classification of online school problems from tweets on Twitter using support vector algorithm

The government has implemented a self-quarantine policy to prevent the spread of the coronavirus. It means prohibiting the gathering of many people. The Ministry of Education and Culture implements an online school policy, so the students still receive an education. At first, people thought that this policy was appropriate and helpful to understand technology better. In any case, people started to complain about almost the different impacts they felt, such as share expenses and stress on online schools. People complain about their problems on social media, Twitter. Hence, it is conceivable to recognize online learning problems through Twitter by categorizing them into two categories, technical and psychological. This scientific research is to classify online school problems to find out the most problem need to be fixed to improve online school quality. The classification of online learning problems uses the text mining method with the Support Vector Machine (SVM) algorithm. SVM algorithm is used to maximized the separation 2 class to decrease errors. The data used in this study were 549 documents with 52% of psychological problems and 48% of technical issues. Using the K-Fold Cross-validation method with K = 10, the average of the training accuracy data is 99.811%, and the average of the testing data accuracy is 90.3%. In addition, the results of the Pres’s Q test show that the model is consistent in predicting the testing data. This research indicates that the Support Vector Machine method is suitable to classify data on online learning problems.