Abstract:
|
In this article, we develop a spectral method for identifying information loss on the process characteristics of an aggregate series. Even though temporal aggregation is a simple and efficient technique summarizing sequential observations, it causes substantial structural changes in a process because a non-aggregate series of a relatively high frequency is transformed into an aggregate series of a relatively low frequency. The effects of temporal aggregation can be explained with changes in the spectral density function. Then, we propose a spectral-based Kolmogorov-Smirnov test for detecting an aggregation resulting significant structural changes to white noise.
|