Practical Predictive Analytics

Tracking down the cause of the outliers

Once you have identified a possible outlier, it is always the best course of action to determine why they occurred. We have illustrated an outlier that appears in one set of data; however, more often than not you will be splitting your data into various subsets to try to track down the cause of the outlier. This is assuming that you have enough data to work with. Outlier detection for small samples is a lot tougher.