Data reduction via stratified sampling for chance constrained optimization with application to flood control planning

Abstract

Due to advanced information technologies, huge data are available to cope with Chance Constrained Problems (CCPs). In this paper, a relaxation problem of CCP is formulated by using such a huge data set. Then a new data reduction method based on stratified sampling is proposed to deal with the huge data set practically. A sample saving technique is also proposed to solve the relaxation problem efficiently by using an adaptive differential evolution algorithm. Finally, the proposed method is applied to the flood control planning formulated as CCP.