Abstract:
In the era of data, the integration of data becomes more and more important. The data may come from different studies, different clients, or the same client but on different aspects. In this work, we are interested in the social platform, where we can observe both the user-user connection data (network) and the user profile/tags/posts (covariates). Our question arises: can we find the “influential covariates”, i.e. covariates that are related to the hidden information of users?
Without network information, this is an unsupervised learning problem. Based on the standard procedure, we propose a network-guided covariate selection algorithm. Leveraging the network information, we significantly improve the selection power. The algorithm is efficient and robust to various network models. Finally, we discuss the downstream applications with selected covariates, including clustering and regression.
About the Speaker:
Dr. Wanjie Wang is an Assistant Professor in the Department of Statistics and Data Science at the National University of Singapore (NUS). She earned her Ph.D. in Statistics from Carnegie Mellon University in 2014. Following her doctoral studies, she completed a two-year postdoctoral fellowship in Statistics and Biostatistics at the University of Pennsylvania before joining NUS in 2016.
Dr. Wang’s research focuses on high-dimensional statistics, social network analysis, and spectral methods, with applications spanning psychology, genetics, and genomics. Her work aims to bridge theoretical advancements with practical solutions to address complex challenges in these interdisciplinary fields.
Your participation is warmly welcomed!
欢迎扫码关注北大统计科学中心公众号，了解更多讲座信息!
