Fig. 2

This chart visually demonstrates how different types of anomalies in a time series exhibit unique signatures in both the time and frequency domains. In the top time-domain plot, we can clearly see a normal periodic signal contrasted with an anomalous signal that contains distinct issues: a sudden spike, a missing seasonal peak, and a segment of high-frequency oscillation. The bottom frequency-domain plot reveals how these anomalies appear after a Fourier Transform. Notably, the spike and the high-frequency oscillation introduce significant energy across the high-frequency spectrum (highlighted in red), creating a stark contrast with the normal signal, whose energy is neatly concentrated at a few low-frequency peaks.