Experimental comparison of representation methods and distance measures for time series data

摘要: The previous decade has brought a remarkable increase of the interest in applications that deal with querying and mining time series data. Many research efforts this context have focused on introducing new representation methods for dimensionality reduction or novel similarity measures underlying In vast majority cases, each individual work particular method made specific claims and, aside from occasional theoretical justifications, provided quantitative experimental observations. However, most part, comparative aspects these experiments were too narrowly demonstrating benefits proposed over some previously introduced ones. order to provide comprehensive validation, we conducted an extensive study re-implementing eight different representations nine their variants, testing effectiveness 38 data sets wide variety application domains. article, give overview techniques present our findings regarding effectiveness. addition providing unified validation existing achievements, also indicate that, certain literature may be unduly optimistic.