Each week, we publish layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format. The article featured today is from the Canadian Journal of Statistics, with the full article in Early View now available to read here. 

Fan, Y., Liu, Y. and Zhu, L. (2021), Optimal subsampling for linear quantile regression models. Can J Statistics. https://doi.org/10.1002/cjs.11590

Big data always contains tremendous amounts of data, which poses serious challenges to its efficient analysis. One solution to this issue is to draw a subsample from the big data and then to make inferences based on the subsample. Although serving the purpose, uniform sampling is usually not the best. To this end, many researchers has developed the best subsampling methods for various models including linear regression, logistic regression, and so on. These methods all require the objective function under study to be smooth enough. However this requirement is violated when we wish to investigate how the median wage is influenced by length of the time employed because the objective function here is not smooth enough. The main contribution of this paper is to develop the best subsampling methods for such problems. The authors find that the new subsampling methods does outperform uniform sampling.