A New GM(1,1) Based on Piecewise Rational Linear/linear Monotonicity-preserving Interpolation Spline

摘要

The classical GM(1,1) model reduces the randomness of the original nonnegative sequence by cumulativelygenerating the original nonnegative sequence to get a monotoneincreasing 1-AGO sequence. The accuracy and usability of themodel will be directly affected by the background constructionmethod. Hence, reconstructing the model background value isof great research significance for improving its prediction accuracy. In this paper, we establish a new GM(1,1) model by usingthe C1continuous monotonic piecewise rational linear/linearinterpolation spline, which provides a more reasonable formulafor calculating the background value. Compared with the classicGM(1,1) model, the new GM(1,1) model has more advantagesin data processing for their effectiveness and accuracy.