Time Series Similarity Analysis using Dimensionality Reduction

In this paper, time series similarity analysis involving dimensionality reduction and fuzzy c-means clustering is applied on time series data between stocks. Derivative Segment Approximation and Von Neumann ratio techniques are carried out for reducing dimensions and similarity can be found by measuring the distances between time series. To quantify the similarity measures, various distance measures viz. Euclidean distance, Bhattacharya distance, Manhattan distance, and Minkowski distance etc., were widely used in analyzing time series data. In particular, this work focused on a Euclidean distance measure in finding the similarity between two time series. The results are reported numerically and graphically.