This article concerns about clustering analysis for curve data with high- dimensional covariates. We propose an adaptive correlation-based approach for achieving clustering. The key idea is to use the shape similarity of structure components to distinguish different clusters from viewpoint of a space basis. Curves with similar shape will be embedding into the same cluster. Our algorithm takes advantage of PDE proposed by Lue (2019) to estimate the loading and basis functions for dimension reduction purpose. Simulation studies illustrate the usefulness of the proposed method.
