A Progressive Sampling based Approach to Reduce Sampling Time

Abstract

Analytics plays vital role in Data Science. It involves finding trends and patterns from the huge repository of data. Scanning huge amount of data consumes lot of time, which can be reduced by sampling. In this paper we have demonstrated effectiveness of Progressive sampling wherein the sample size is gradually increased till it reaches a desired accuracy. By applying an algorithm based on Rademacher average to mine frequent datasets using Progressive sampling, we have shown that the runtime and the sampling time is considerably reduced as compared with static sampling.