Real time interpretation and optimization of time series data stream in big data

摘要

In view of massive historical data and high-speed time series data stream in the futures market program trading, how to mine related data and explain data real-timely, this paper proposes a real-timely proceeding model for massive data based on ARTMMR algorithm. The ARTMMR algorithm obtains frequent itemsets by using the parallelism of MapReduce, which saves the storage space and decreases the time overhead. The CPU utilization is improved by using Batch algorithm and thread calling algorithm, it meets real-time processing requirement and excavate trading opportunities or feature model according to the requirements of traders. The results indicate that the model can not only explain the time series data stream real-timely, but also help traders analyze data quickly and achieve accuracy trade-offs.