Univariate Outlier Detection

Anomaly detection is simplest in the univariate case, that is, when each observation is only one number. In this case, we might start by doing a common-sense check for anomalies by checking whether observations are missing, NULL, NA, or recorded as infinity or something that doesn't match the type of the rest of the observations. After performing this check, we can apply true unsupervised learning.

For univariate data, anomaly detection consists of looking for outliers. R's built-in boxplot function makes an initial exploratory check for outliers quite easy, as can be seen in the following exercise.

Exercise 37: Performing an Exploratory Visual Check for Outliers Using R's boxplot Function

For univariate data, anomaly detection consists of looking for outliers. R's built-in boxplot function makes an initial exploratory check for outliers quite easy, as demonstrated in this exercise. We will use a dataset called mtcars, which is built into R.

In this exercise, we...