Neither the algorithm of symbolic aggregate approximation (sax) nor the symbolic algorithm for time series data based on statistic feature (sfvs) will involve in the shape of the time series, so it cannot effectively represent the similarity of the time series. In this paper, a symbolic method for time series based on mean and slope is introduced to represent the similarity of the time series. It firstly, segments the time series based on key points, then symbolizes the mean and slope separately, records every symbol's occurrence times and position, finally uses every symbol's occurrence times and position as the metrics standard. The experiments show that this method can be used effectively for time series similarity matching, and also improve the correct rate.
展开▼
