Outlier data optimal identifying and detection of based on chaotic
Jianzhou Wang · Ha'erbin gongye daxue xuebao · 2006
Identification of outliers can lead to the discovery of unexpected,useful knowledge,and has lots of applications.A novel approach is proposed for outliers detection and mining.We can compute the slope between two data points first in time series,and compare the slope with chaotic forecasting slope to confirm outlier set,then prove the optimality of the mathematics model,and evaluate the loss of mathematics mode information.