Hi @rick-wheller ,
Can you provide more details about your use case? We have multiple unsupervised methods to detect anomalies built into the platform. Also, many of our blueprints use unsupervised methods (PCA, k-means clustering, etc.) to generate additional information for supervised models (e.g. XGBoost).
@rick-wheller No problem! We have isolation forest, one class SVM, and local outlier factor, and some other statistical techniques available.
One quick note: Isolation Forest and double median will perform the fastest on larger datasets.
I am unable to find a FREE course on DataRobot University on this topic. Can you please direct me to where I can find a tutorial on Unsupervised learning , clustering methods and artifacts related to this, basic intro before I can learn about Anomaly detection.
Hi @ssukavan - Have you had a change to look at the information @dalilaB shared? Does this have what you're looking for? If so, please help others find the best answer by selecting her response as the solution. If not -- what more do you need? I'm sure other members would love to help you out.
Let us know!
Its great that I can see documentation on the topic I requested , but dont see the option in my installation. Is it a version issue or a cloud/on-prem issue. Ours is an on-prem installation.
Yes, in 7.2 we had anomaly detection in AutoML and in AutoTS. However, clustering was in preview. If you want to perform anomaly detection just make sure you don't fill the target field and click on no target. Also, please find the link to the release notes here