Multivariate functional subspace classification for high-dimensional longitudinal data

We propose a multi-class classification method for multivariate functional data using the subspace method. The subspace method reduces the dimension of data for each class by mapping the data onto the subspaces, and then classifies the data to the class with the largest similarity to the subspace. We apply multivariate functional principal component analysis to the data for each class to obtain the subspaces. Since the subspace based on the functional data depends on time, we integrate the similarity between the data to be classified and the subspace. The proposed method can be easily applied to high-dimensional data, since it greatly reduces the dimension of the data. Simulation and real data analysis show that our method provides effective classification results.