Types of Anomalies in Anomaly/Outlier Detection

There are three types of anomalies or outliers in quantitative analysis: global, contextual, and collective anomalies.

1. Global anomalies refer to instances where certain data points are significantly away from the rest of the data points. For example, if a person makes a purchase of $5,000 while their average spending is $100, it can be suggestive of an anomaly worth checking.
2. Contextual anomalies consist of instances in which certain data points are considered to significantly deviate from the rest of the data under specific conditions. The deviation depends on contextual information. For example, a certain temperature reading may be deemed too high in one region but normal in another area.  In that particular case, thus, temperature data points should be judged too high or otherwise in a context-specific manner.
3. Collective anomalies indicate cases where certain closely grouped data points deviate from the rest of the data points; that is, they deviate as a group. For example, if the power suddenly goes out in a certain block of houses, it can be thought of as a collective anomaly since it is unusual for all the houses to go dark all at once.