The rapid advance in science and technology in the past decade brings an extraordinary amount of data that were inaccessible just a decade ago, offering researchers an unprecedented opportunity to tackle much larger and more complex research challenges. The opportunity, however, has not yet been fully utilized, because effective and efficient statistical and computing tools for analyzing super-large dataset are still lacking. One major challenge is that the advance of computing technologies still lags far behind the exponential growth of database. One option is to invent algorithms that make better use of a fixed amount of computing power.

In this talk, I will review an emerging family of subsampling methods that are developed for achieving such a goal.  In subsampling methods, we sample a small proportion of the data (subsample) from the full sample, and then perform intended computations for the full sample using the small subsample as a surrogate.  In classic statistical literature, subsampling has been used to refer to ‘m-out-of-n’ bootstrap, whose primary motivation is to make approximate inference owing to the difficulty or intractability in deriving analytical expressions.  The general motivation of the subsampling methods in data science is different from the traditional subsampling. I will present challenges and opportunities.