Personality Prediction of Social Network Users

Abstract

Through weibo users, we extract social data and questionnaire, and focus on how to use the user text information to predict their personality characteristics. We use the correlation analysis and principal component analysis to select the user information, and then use the multiple regression model, the gray prediction model and the multitasking model to predict and analyze the results. It is found that MAE values of the gray prediction are better than the multiple regression model Multi-task model, the overall effect of the prediction between 0.8 and 0.9, the overall accuracy of good prediction. This shows that gray prediction in the user's personality prediction shows a good generalization and non-linear ability.