Assessing Strength of Seasonality Through Sample Entropy: A Simulation Study

This paper investigates the behaviour of sample entropy when used as a measure of seasonality of time series. Sample entropy decreases when the series becomes less complex or when regular patterns emerge. The more regular patterns in seasonal data compared to those of non-seasonal data is used in providing evidence that sample entropy is inversely related to the likelihood that seasonality exists in the data. A simulation study was conducted to assess the behaviour of the sample entropy in relation to seasonality. Sample entropy yields large values for time series without seasonality, and as the extent of seasonality becomes dominant, the value decreases. The sample entropy becomes a more reliable measure of seasonality as the length of the time series increases.