Anomaly Detection Using ML.Net

Swapnil Dave 41

I want to detect anomaly for the time series i am having and i want to understand how does the microsoft detects anomaly withe the different algorithms it provides.

The problem or the challenge i am facing is that the method arguments. I don't know what arguments should i pass in the algorithm.

Like For Example:

DetectAnomalyBySrCnn has arguments like :- trainingWindowSize, threshold, backAddWindowSize, and many more.

How to pass these parameters in a way that i get correct output.

1 answer

Ramr-msft 17,736 Reputation points

2021-09-24T14:30:11.923+00:00

@Swapnil Dave Thanks for the question. Can you please share the sample/Document that you are trying.
SR in ML.NET and SR in Python: https://github.com/microsoft/anomalydetector
Try out the service at Azure Notebook: https://aka.ms/adnotebook
Here is link to the document using ML.NET.
Please sign in to rate this answer.
Swapnil Dave 41 Reputation points

2021-09-25T06:41:39.167+00:00

135097-srcnn.txt

@Ramr-msft
This is the file pretty much the gist of what i am trying to do..

Actually i am a fresher in this ML domain and trying really hard to gain knowledge. I have the task of detecting anomaly in the time series. I have been told that there are two types of anomaly detection, namely : 1] basic 2] agile (like seasonal maybe)

And, also i have been told about that ML Dotnet is good for that. And i have been trying to get a grasp on the way ML code is done in c# dotnet and all the algorithm it provides. I have seen the algorithms for the anomaly detection but i can't understand what these algorithms perform underneath. As of now the computations are just a black box to me, but if provided the right understanding i might be able to perform well. I understood the pipeline flow of ML Context and all but the thing that concerns me is the working of the algorithm. And if i don't know how the algorithm performs i won't be able to pass the required arguments. In a way that they are intended to.

I would love to learn and gain the knowledge about the algorithms if you could teach me. I don't have any teacher who could teach me this.

P.S: I have passed random arguments in the SrCnn Anomaly model for the windows size, training size, threshold and all. Was just checking
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

Anomaly Detection Using ML.Net

1 answer

Your answer