Time Series:Similarity Search and its Applications

A model for identifying similar time series has been developed. Two time series are considered similar if they have enough non overlapping time ordered subsequences that are similar. The two subsequences are considered to be similar if one is enclosed within an envelope of a user defined width around another. Amplitude of one of the two sequences is scaled by a suitable amount to make the matching scale invariant. We also allow a small gap between the matching subsequences. A static index structure namely kdtree is used to index the time series and the Range search methodology is used to search for the similar subsequences. The highlight of this model is that the similar subsequences are not stored repeatedly in the index structure, but only the time stamp corresponding to it is updated in the external node of the kd-tree.Interesting results are found to show the validity of the model.