Regularized extreme learning machine ensemble using bagging for tropical cyclone tracks prediction

Abstract

This paper aims to improve the prediction accuracy of Tropical Cyclone Tracks (TCTs) over the South China Sea (SCS) and its coastal regions with 24 h lead time. The model proposed in this paper is a regularized extreme learning machine (ELM) ensemble using bagging. A new method is proposed in this paper to solve lasso and elastic net problem in ELM, which turns the original problem into familiar quadratic programming (QP) problem. The forecast error of TCTs data set is the distance between real position and forecast position. Compared with the stepwise regression method widely used in TCTs, 16.49 km accuracy improvement is obtained by our model. Results show that the regularized ELM ensemble using bagging has a better generalization capactity on TCTs data set.