Time Series Clustering Based on Singularity

Abstract

With relevant theories on time series clustering, the thesis makes research into similarity clustering process of time series from the perspective of singularity and proposes the time series clustering based on singularity applying K-means and DBScan clustering algorithms according to the shortage of traditional clustering algorithm. In accordance with the general clustering process of time series, time series clustering based on singularity and K-means are made respectively to get different clustering results and make a comparison, thus proving that similarity clustering research of time series from the perspective of singularity can better find out people's concern on time series.