A spectral-based approach for clustering of multivariate time series
碩士 === 國立臺北大學 === 統計學系 === 100 === The application of spectral analysis in time series clustering can overcome the problem of starting point or different scaling, and it can enhance the data reduction. However, the relative researches for multivariate time series clustering are rare, we propose the...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2012
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/02712332477362275365 |
| Summary: | 碩士 === 國立臺北大學 === 統計學系 === 100 === The application of spectral analysis in time series clustering can overcome the problem of starting point or different scaling, and it can enhance the data reduction. However, the relative researches for multivariate time series clustering are rare, we propose the multi-dimension Linear Predictive Coding (LPC) cepstrum by using the Euclidean distance between two time series as their dissimilarity measure for time series clustering.
In the simulations , we find that multivariate Linear Predictive Coding clustering perform better than other multivariate spectral clustering beyond on the simulated or silhouette criteria. We also apply this method for the international stock market clustering.
|
|---|