Age prediction based on feature selection

Microblog as one of the most popular social network sites, has become an important platform for people to share, exchange, obtain and spread information in their life. This paper infers the Microblog user's age range from the perspective of feature selection. We analyze the user microblog text from six kinds of features. This paper has two distinctive characteristics: (i) a new feature selection method is proposed to improve the existing method based on word frequency. (ii) we add time features that users post microblog. This is a new perspective to analyze the user's age. Experiments are conducted on these features and demonstrate that the features we proposed outperform previous features. Finally, visualization can help to understand and analyze the results of the inference, but also has application value.