Detection of Anomalies in Water Networks by Functional Data Analysis

Figure 10

The left-hand panel shows the outlier detected by FB in the first iteration, while those detected in the second iteration are shown in the right-hand panel. The legends indicate the outliers. They belong to the group of weekends in spring for sector B. Time is expressed in hours and flow in cubic meter per hour (). The procedure for outlier detection in Stage 2 is not restricted to FOADA; this is the reason why we have used, for instance, FB to illustrate this.