Time series anomaly detection

@ Founded in ：2022.07.25
@ Modified to ：2022.07.25

1、 Method

（1） Business interpretation 、 Statistics （3 $\sigma$, boxplot ）
（2） Consider it as a clustering problem （kNN,kMeans）
（3） See it as a prediction problem
Time series prediction

（4） After temporal decomposition , Statistical residual or fitting residual
（5）Isolation Forest
（6） Sliding window method （ Left and right sliding windows , Expansion sliding window ） And coefficient of variation
（7） Deep learning method

2、 Tools

Featuretools
pyod

3、 Value connectivity

Time series anomaly detection （ One ）—— Algorithm overview
Time series anomaly detection （ Two ）—— be based on KDD99 The actual combat of data sets
Time series anomaly detection （ 3、 ... and ）—— Alarm strategy and root cause location
Time series anomaly detection （ Four ）—— Root cause positioning landing practice