Mining periodic patterns in manufacturing test data

Abstract

Mining of periodic patterns in time series databases is an important data mining problem with many applications. Previous articles have considered the mining of periodic patterns in datasets that range from standard market basket datasets to datasets containing information about the movement activities of cellular phone users. Each of these studies offer solutions to the given domain but lack the ability to address the domain of manufacturing test data. This paper proposes a general model for discovery of periodic patterns within datasets related to the manufacturing of electronic goods. Three general phases are considered. The discretization of the original dataset is first to be discussed, followed by the clustering of the dataset into state related clusters and finally the discovery of periodic patterns in the state transitions of the tests.