Discovering hierarchical index structures in time series archives
Hassan Ali Artail · 2003
We describe an algorithm that uses feature detection as the basis for discovering index parameters within temporal data archives that comprise multiple time series. The detection of features is accomplished using the Haar wavelet transform. We focus on the discontinuities whose number and location within each series are used to extract index information and then build a hierarchy of index vectors. We present a client application that applies the algorithm to automotive test data archives.