The comparison of data mining structure performance for depressive classification from posts on twitter of user behaviors

Abstract

In 2018, The World Health Organization (WHO), specified that the major depressive disorder (MDD) was the second disease that it is probably cause of Social media usage affect to stress and violence, and lead to depression. This research proposed the Depressive classification from posts on twitter of user behaviors and compared the classifier between 1 level and 2 levels: 1) 1 level: using Bayes algorithm created a model for classification between general and symptoms based on a symptoms of questionnaire (DSM-5) including as follow: 1) depressive 2) loss of interest 3) appetite 4) abnormal sleep 5) slowed thinking 6) guilt 7) tired 8) unexplained and 9) suicidal ideation. 2) 2 levels: 2.1 Using SVM algorithm created a model for classification between general and depression. 2.2 Using Bayes algorithm compared by Random Forest algorithm for classification a symptoms of questionnaire (DSM-5). The data came from real tweets of International celebrities set is divided into 2 sets: training set and test set. Finally, the results are demonstrated of training set prediction between 1 level and 2 levels: 1. 1 level: Bayes algorithm showed that the accuracy = 82.55%. 2. 2 level: 1) SVM algorithm showed that the accuracy = 98.20%. 2) SVM algorithm pair with Bayes algorithm showed that the accuracy = 82.23%, and SVM algorithm pair with Random Forest algorithm showed that the accuracy = 91.45%. The results of test set, by the boundary of probability are variously set 0.1 to 0.9 that prediction between 1 level and 2 levels: 1. 1 level: Bayes algorithm showed that the accuracy = 76.67%. 2. 2 level: 1) SVM algorithm pair with Bayes algorithm showed that the accuracy = 73.33%. 2) SVM algorithm pair with Random Forest algorithm showed that the accuracy = 70.00%.