The Improved SVM Method for Forecasting the Fluctuation of International Crude Oil Price

摘要

Forecasting the fluctuation of international crude oil price has been the major focus of economics due to recent drastic fluctuation of international crude oil price. In this article, we forecast crude oil price at a daily frequency based on a classification techniques: cluster support vector machines (ClusterSVM). We improved ClusterSVM by exploiting the distributional properties of training data and accelerated the training process with large-scale data set. The algorithm partition the training data into disjoint clusters, then train an initial SVM using representatives of these clusters. Based on initial SVM we can approximately identify the support vectors and non-support vectors. The training process is accelerated by replacing non-support vectors with few data. The initial support vectors of cluster are the key of training ClusterSVM. The improved ClusterSVM can obtain the initial support vectors efficiently. Experiment results indicate that the improved ClusterSVM method excel conventional SVM method for forecasting fluctuation of international crude oil price.