Predictive modeling of large-scale curves and its application on GDP prediction of multi-regions

Traditional approach to predict large-scale sequential curves is to build model separately according to every curve, which causes heavy and complicated modeling work inevitably. Therefore the existing approach is lack of manipuility in the application. A new method is proposed in this paper to solve this problem. By reducing model types of curves, clustering curves and modeling by clusters, the new method simplifies modeling work to a large extent and reserves original information as possible in the meantime. This paper specifies the theory and algorithm, and applies it to predict GDP curves of multi-region, which confirms practicability and validity of the presented approach.