Enhancement of Non-stationary Time-Series Clustering Analysis Using Projection Pursuit Regression Method
Tsair-Chuan Lin · 2010
The theory and application of time-series clusterin g analysis is an effective explanatory technique in various research fields. To overcome the limitations and ma ny assumptions in conventional model-based clusteri ng, this study utilizes the projection pursuit regression method a s explanatory tool for formulating, identifying and estimating nonlinear models to approach the complex regression surface and then applied the projection pursuit me thod and an agglomerative scheme to cluster time series based o n a similarity measure. This clustering can be appl ied to nonlinear, non-stationary, non-Gaussian models, and models inv olving interactions in predictor variables. Simulat ion results and real data analysis for categorizing the collection of average personal income of 25 states in the US d emonstrate that this scheme compares favorably with other methods f or similar clustering tasks.