Hi Soren,

anomaly detection being a one classifier problem, you can pick up data sets from machine learning repository from UCI and begin with simple exploratory data plots and clustering. So a very simple example would be say to take a breast cancer data set, apply PCA and get it on to two dimensions, begin with K-means clustering to get the mean of the data set and then build a robust outlier detection model over it, this would be a very naive idea for an anomaly detection system. Hope this solves your query.

Hello Soren,

Please go through this presentation with examples,

Please, see article

Enabling the Discovery of Recurring Anomalies in Aerospace

Problem Reports using High-Dimensional Clustering

Techniques

Ashok N. Srivastava, Ph.D., NASA Ames Research Center, [email protected],

Ram Akella, Ph.D., University of California, Santa Cruz, [email protected],

Vesselin Diev, University of California, Santa Cruz, [email protected],

Sakthi Preethi Kumaresan, University of California, Santa Cruz, [email protected],

Dawn M. McIntosh, NASA Ames Research Center, [email protected],

Emmanuel D. Pontikakis, University of California, Santa Cruz, [email protected]

Zuobing Xu, University of California, Santa Cruz, [email protected],

Yi Zhang, Ph.D., University of California, Santa Cruz, [email protected]

2006 IEEE Aerospace Conference Proc.