Large-scale Time Series Data Down-sampling Based On Map-Reduce Programming Mode
In the last decades,more and more time series data has been collected in many kinds of fields,and specially in the industry field,which has been increased greatly.One of the most common types of data visualization used is the line chart,but in industry field,time series datasets are so huge that it costs much more time to draw data as a line chart.In this case,we must reduce dimensionality of time series data to keep the features of the raw data.There are some data sampling methods to reduce data,such as Discrete Fourier Transform(DFT),Discrete Wavelet Transform(DWT),Singular Value Decomposition(SVD)and Largest Triangle etc.However,these methods will cost much time for data reduction in that time series data sets are very huge.In this paper,we will introduce time series data down-sampling algorithms and Map-Reduce(MR)programming model,and propose a time series data down-sampling method for large-scare time series data based on Map-Reduce programming model.At last,we will show the result of the methods based on the time series data sets collected from sensors.
large-scale data Time series data down-sampling Map-Reduce
Jiajia Xu Yichang Qiu Haiying Zhang Meng Li Manli Li
Beijing Institute of Spacecraft Environment Engineering
国际会议
重庆
英文
409-413
2017-03-25（万方平台首次上网日期，不代表论文的发表时间）