Summary
The important takeaway of this chapter is about being observant but not being paranoid. An anomaly is always unexpected, but it’s not always malicious or an indicator of impending doom. Some anomalies are actually welcome because they’re novelties that signify a trend toward something positive. The techniques that this chapter contains help you to differentiate between novelties and hacker attacks so that you don’t waste time chasing data that doesn’t matter in security matters.
A large part of this chapter focused on showing various techniques for discovering anomalies so that you can mitigate them. Even though the univariate approach may seem weak, it also has the benefit of being both fast and simple. You should first try the univariate approach before moving on to the more complex techniques used for multivariate analysis. When it comes to security, speed and simplicity do matter, and some advice you might find in data science texts for fully...
