Edgeworth Expansion for Summation of Symmetric Statistics and Its Application in Analyzing Massive Data

• Activity Number: 656
• Type: Contributed
• Date/Time: Thursday, August 4, 2016 : 8:30 AM to 10:20 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Liuhua Peng* and Song Xi Chen
• Companies: Iowa State University and Peking University/Iowa State University
• Keywords: Edgeworth expansion ; symmetric statistics ; massive data ; split-and-conquer ; high order correctness ; computational efficient

Abstract

We proved Edgeworth expansion for summation of symmetric statistics to the third order, and this can be applied to analyzing massive data set by split-and-conquer, both estimator and confidence interval are derived. Our theoretical results guarantee the high order correctness of our proposed method. Moreover, it's more computational efficient by doing split-and-conquer.