Clustering of longitudinal curves via a penalized method and EM algorithm
share
In this article, the subgroup analysis is considered for longitudinal curves under the framework of functional principal component analysis. The mean functions of different curves are assumed to be in different groups but share the same covariance structure. The mean functions are written as B-spline functions and the subgroups are found through a concave pairwise fusion method. The EM algorithm and the alternating direction method of multiplier algorithm (ADMM) are combined to estimate the group structure, mean functions and covariance function simultaneously. In the simulation study, the performance of the proposed method is compared with the existing subgrouping method, which ignores the covariance structure, in terms of the accuracy for estimating the number of subgroups and mean functions. The results suggest that ignoring covariance structure will have a great effect on the performance of estimating the number of groups and estimating accuracy. Including pairwise weights in the pairwise penalty functions is also explored in a spatial lattice setting to take consideration of the spatial information. The results show that incorporating spatial weights will improve the performance.
READ FULL TEXT