2.4 Very extreme data values

An outlier is a data value that is noticeably separated from, or inconsistent with, the rest of the data.

Suppose a rainfall value for Camborne in July 2023 was accidentally input incorrectly and a value of 313mm was recorded instead of 113mm. This value would then be noticeably separated from the rest of the data, so it would be an outlier. Here is how you would represent this data set on a boxplot:

Figure 8 Box plot of rainfall, Camborne

An asterisk is used to represent the data value that falls a significant distance away from the rest of the data.

Outliers can also be called anomalies, meaning they are outside what is considered normal. Most real data sets contain anomalies. Data scientists use various strategies for identifying and removing them, depending on the statistical techniques they intend to use and how important it is to recognise extreme events. This 313mm data value may be considered a mistake and could be corrected or deleted. Other anomalies may be genuinely valid data values that seem very extreme when compared to the rest of the data. Anomalies that result from rare, unexpected events are important in identifying system flaws or failures such as fraud or breakdown in a model.