B-spline inspired multivariate grey model for short-term time series forecasting

The multivariate grey model (MGM), which is recently improved by virtue of the convolution integral, has emerged as a powerful tool for the prediction problem. Unfortunately, this promising technique only effectively adopted the trapezoidal rule, whereas the model coefficients and other interpolation methods are not fully considered. In this paper, we propose an alternative version of MGM inspired by B-spline. Focusing on the evaluation of the model coefficients and convolution integrals, which are key elements for improving the efficiency of MGM, we replaced the existing trapezoidal rule with B-spline. Consequently, comparative studies of the proposed schemes and other generally acknowledged methods are conducted on synthetic data. Simulation results indicate that the proposed methods can achieve promising reinforcement in the short-term time series forecasting performance.