Time series data are a set of data high-frequently collected in chronological order and the huge amount causes difficulty in observation and utilization. Thus, we present a method to rapidly generate and visualize the digest of massive time series data. The method comprises the following steps: linearly scanning the input time series data, dividing the time series data into a plurality of data intervals, calculating the statistical data, characterizing the data features in each interval, and adopting an improved candlestick chart to show the trend and distribution of the data series. This method is less complex with small amount of data, and can store and query data in relational database. The improved visualization method of candlestick charts is more practical and operable, which can visualize the distribution of data through color change showed in the candlestick charts. During the visualizing process, it can support the merging and separating of the intervals at different levels and granularity. Through experiments, it is proved that this method is efficient and useful.
[1]
Luo Yang.
Fault diagnosis based on sequential pattern mining
,
2010
.
[2]
Tak-chung Fu,et al.
Flexible time series pattern matching based on perceptually important points
,
2001
.
[3]
Li Ai.
On-Line Segmentation of Time-Series Data
,
2004
.
[4]
Deng Wei.
HMM-based hierarchical clustering of gene expression time series data
,
2011
.
[5]
Dai Qi.
Time Series Segmentation Based on Fixed Number of PIPs Detection
,
2011
.
[6]
Mu Zhong-lin.
Time series piecewise linear representation based on trend transition point
,
2010
.
[7]
Pedro V. Sander,et al.
Signal-specialized parameterization for piecewise linear reconstruction
,
2004,
SGP '04.