Machine learning classification model using Weibo users' social appearance anxiety

Social platforms aggravate social appearance anxiety (SAA). Therefore, identifying groups with high levels of SAA on social media is critical. Psychological indicator classification and modelling by using social media data can be performed without intrusion. Furthermore, accurate psychological portrayal of social desirability can be obtained using social media data. The study extracted 9 Weibo features related to SAA based on theoretical basis. A support vector machine was used to establish a relationship between the Weibo user data and SAA scale. The results revealed that the accuracy (ACC) of using the activity history of Weibo users to identify SAA was approximately 73.8%. The high-ACC automatic classification of users' SAA can be directly accomplished by analysing users' social media behaviour data. The results of the study can be used to distinguish for SAA and help users develop a positive and reasonable body image. •A social appearance anxiety classification model is established.•A SVM classification model is used to identify the SAA level of Weibo users.•The mapping relationship between the SAA scale and feature vector is established.•Nine features of Weibo improve the prediction performance of the model.