Qfyilyi

I have set of time series data points having 4 variables say (A,B,C,time). I am working on finding anomaly detection in the data.

I found tools which find anomalies in 2D data e.g. Yahoo EGADS library for finding anomaly detection. Meaning I can provide this tool (A,time) or (B,time) input etc and it will find anomaly detection in that.

Now the problem is that I need to find anomaly detection for all A,B pairs vs time. A,B pairs can be large. It seems inefficient to run time series models for all A,B pairs vs time parallely.

Can anyone suggest some way using this library or some other library which can solve the purpose.

There are many libraries in R and python. There are some solution given in R-

what is anomaly?-
https://machinelearningstories.blogspot.com/2018/07/anomaly-detection-anomaly-detection-by.html

Connectivity based outlier technique-
https://machinelearningstories.blogspot.com/2018/09/connectivity-based-outlier-detection.html

For high dimensional data-
https://machinelearningstories.blogspot.com/2018/08/anomaly-detection-in-high-dimensional.html

Even QQ plot of data can give you abnormality. These are unsupervised algos, so even after finding outlier( anomaly) , you need to relate it with actual abnormality.

There are many packages in Python also