Finding deviations in your data with outlier detection

In data analysis, uncovering meaningful insights often includes identifying patterns, trends, deviations, and anomalies. One useful technique is outlier detection—it involves detecting data points that significantly deviate from the majority of the dataset. This helps us identify elements that differentiate normal data points from anomalous ones. Unlike anomaly detection for time series data, as we learned in the previous chapter, we are not concerned with the temporal evolution of the dataset. Instead, we focus on data clusters, evaluating their density and distance using multivariate analysis.

In the first four recipes of this chapter, we will continue to use the Rennes traffic data, which has become familiar to us from previous chapters.

Getting ready

Before delving into the supervised learning intricacies in the context of the Elastic Stack, let’s review the ML methodology from end to end, as shown in...