• Thanks for the article. So how you define an outlier in an automated process? Those observations that fall in the first cluster?