Discriminant and cluster analysis of possibly high-dimensional time series data by a class of disparities
英明 長幡 · Institutional Repositories DataBase (IRDB) · 2017
For possibly high-dimensional time series data, a basic discriminant statistic has a goodness, and this can be applied to a classification of companies.Discriminant and cluster analysis of high-dimensional time series data have been an urgent need in more and more academic fields.For possibly high-dimensional and stationary time series data, we show the consistency of classifier under suitable conditions.Also, simulation studies show that that works even in the case of finite observations of training samples.Finally, we conduct the cluster analysis for real financial data.We conclude that our method is suitable for the discriminant and cluster analysis of high-dimensional dependent data.