Title: Distributed computing for quantile regression: A statistical analysis
Authors: Shih-Kang Chao - Purdue University (United States)
Guang Cheng - Purdue Univ (United States)
Stanislav Volgushev - University of Toronto (Canada) [presenting]
Abstract: With emergence of new data collection and storage technologies, it has become easy to accumulate extremely large data sets. At the same time, statistical analysis of such data poses serious computational challenges. One common approach to handling the resulting computational burden relies on splitting the complete data set into smaller subsamples and performing computation on each of the subsamples. While such an approach is easy to implement, the theoretical properties of resulting procedures remain largely unclear. We provide a detailed analysis of such a splitting approach to quantile regression and discuss potential applications, including the analysis of panel data models.