ENSEMBLE-BASED TIME SERIES DATA CLUSTERING FOR HIGH DIMENSIONAL DATA
Sampasetty Saravanan, Gulam Mohideen Kadhar Nawaz · 2014
The time series clustering analysis provides an effective way to discover the intrinsic structure. In most of the time, the series of data mining algorithms uses similarity search as the core subroutine, and hence the time taken for similarity search becomes complicated, due to the large data sets. In this paper, we have developed an approach for clustering the temporal data via the ensemble of cluster weight for multiple partitions developed by initial clustering analysis on two types of representations. Ini- tially, time series data sets are converted into representations in which each partition is used to reduce the dimension and subsequently, the clustering algorithm is applied. The different types of weight algorithms are applied to each of the representation. By consid- ering the weight and the representation matrix, we develop thenal clustering. Finally the experimentations are carried out on the time series data sets, and the simulation results demonstrate that our approach gives the desired results in clustering analysis of time series data. Keywords: Representations, Time series data, Representation clustered matrix, Wei- ghted consensus function, Fuzzy- C-means (FCM), Kernel function, Temporal data clus- tering