Data Anomaly Detection through Semisupervised Learning Aided by Customised Data Augmentation Techniques

Figure 8

Precision/recall curve of each pattern with different DA compositions: (a) normal, (b) missing, (c) minor, (d) outlier, (e) square, (f) trend, and (g) drift.

(a)

(b)

(c)

(d)

(e)

(f)

(g)