A DWT based time series outlier data mining algorithm

Outlier data mining can often find some potential useful information, this article proposes a outlier mining algorithm for time series data. The algorithm first uses DWT (Discrete Wavelet Transform) to transform the time series data to frequency domain from time domain and makes the time series data map into the points in multidimensional space. Then the algorithm uses distanced-based method to mine the outlier data. Simulation result shows the effectiveness of the algorithm. The algorithm in this paper need not to know the probability distribution model of data points and different function, thus it overcomes the limitations of previous methods of data mining.