front matter
preface
This book is the result of many years of thinking about, and becoming fascinated by, the problem of outlier detection. I’ve worked in a number of jobs where outlier detection was a practical necessity of the work—for example examining financial data and performing social media analysis. So, there has been a very practical element to learning outlier detection. But outlier detection was also likely the most intellectually interesting area I worked on, even after many years of working in numerous other areas of data science.
The general idea of outlier detection is fairly simple: finding the items in a dataset that are most unlike the others. But, in practice, it’s often quite difficult to do in an efficient and effective way, particularly where there are subtle outliers you’re interested in. Once executed, it’s difficult to determine if the items flagged as the most anomalous truly are the most anomalous. In fact, it can be difficult to even identify specifically why the items flagged are anomalous.