I'd like to develop an anomaly detection. I have historical data from sensors in the form of time series. The time series can be divided into data of a normal state and data of an abnormal state i.e. I have good data and bad data from sensors. Before I want to implement the actual anomaly detection, I want to investigate whether the good data and bad data can be separated by an explorative analysis. Thereby I want to find out if I can distinguish between good data and bad data. I hope to find out if anomalies can be detected later. Now I am looking for statistical methods or algorithmic procedures to detect and visualize this.

Can anyone help me or has ideas how I can analyse this?


  • $\begingroup$ The bible on this thread starts here docplayer.net/… and continues here faculty.chicagobooth.edu/ruey.tsay/teaching/uts/lec10-08.pdf and here stats.stackexchange.com/questions/169468/… $\endgroup$
    – IrishStat
    Commented Apr 9, 2019 at 21:11
  • $\begingroup$ @Irish I'm having a hard time seeing the applicability of your methods to the present question: it concerns a situation where there appears to be a potential to characterize the anomalous patterns, perhaps quite precisely. One needn't search for just any "intervention," but can focus--to great advantage--on detecting the targeted patterns. Makome, it would help to clarify your question concerning the nature of these anomalies and your objectives. For instance, would this be post hoc analysis or would you need to detect a new occurrence of an anomaly as quickly as possible in real time? $\endgroup$
    – whuber
    Commented Apr 9, 2019 at 22:08
  • $\begingroup$ what i inferred ( perhaps mistakingly) was that normal observations were being observed and then at subsequent chronological periods potentially unusual observations were recorded. The mission was to red-flag them as soon as they occurred. $\endgroup$
    – IrishStat
    Commented Apr 9, 2019 at 23:07


Browse other questions tagged or ask your own question.