MAP REDUCE FOR R/S ANALYSIS OF HIGH-DIMENSIONAL TIME SERIES

Sabutkevich A. M., Vikhlyaev D. A., Nikiforov I. V.

The work is devoted to research approaches to improve the efficiency of computing indicators of high-dimensional time series, presented using ungrouped streaming data, using the example of using R/S analysis. The proposed method is based on the use of the distributed computing model Map Reduce for the implementation of the R/S analysis algorithm. The proposed solution is implemented in the software tool, which makes it possible to increase the efficiency of calculations using the correct cluster configuration by an average of 34% compared to the traditional computing method for experimental dataset.

Keywords:distributed computing, Big Data, Map Reduce, R/S analysis, Hurst exponent, time series