Data Mining and Knowledge Engineering

Kaushik Raj Palanichamy ¹, Selvakumar Subbiah ²

Affiliations
1 Cognizant Technology Solutions, Coimbatore, Tamil Nadu, IN
2 Department of Statistics, St. Joseph College, Tiruchirappalli, Tamil Nadu, IN

Abstract

Time series is a sequence of values observed over the time. There are several patterns such as periodic patterns, similarity patterns, seasonal patterns, etc. This paper deals with the similarity analysis, which is concerned with efficiently locating subsequences in large archives of sequences. It also discusses the appropriate use of Piecewise Constant Approximation (PCA) and coefficient of variation method for data reduction technique. Finally, the fuzzy c-means and k-medoid cluster analysis are applied to the reduced data to measure the similarity between two sequences and the results are compared numerically and graphically.

Keywords

Similarity Search, Data Reduction, Coefficient of Variation, Distance Measures, Clustering, Fuzzy C-Means and K-Medoids.

Full Text