The Improvements Of GMDH Algorithm And Its Application In Coal Price System

The coal energy is an important means of production which related to China's national economy and people's livelihood, and plays an active and important role in the national economic construction. Coal price is a critical factor in the coal market, the fluctuation of which not only affects the supply and demand balance in the coal market, but also adjusts production and operating activities of supply and requisitioning parties. It can provide the basis for coal enterprise management decision-making and national economic policy. Just under such a background, doing research about forecast of coal price system has a very important practical significance.This paper tries to use GMDH algorithm to do some research of the modeling and prediction of the coal price system regarded as the research object of this paper. First, the main factors which influence coal price in the coal price system are analyzed qualitatively, the coal price mechanism both at home and abroad are also introduced. And then this paper represents the principle and modeling process of GMDH algorithm, synthesizes advantages and defects of classical GMDH algorithm. On account of three main defects analyzed: the sample classification method is artificial, use regularity criterion as the external criterion can not guarantee the extrapolation ability of the model, and the OLS method has its own deficient, the GCG-GMDH algorithm is proposed, which presents relevant improved strategies: divide the sample data by group, use combined criterions instead of single criterion, and introduces the GA algorithm to take the place of OLS method. The classical GMDH algorithm and GCG-GMDH algorithm are adopted to bulid the coal price system respectively, and on this basis, the coal prices in the next three years are forecasted. By comparison, the GCG-GMDH model behaves better than the classical GMDH model in prediction accuracy.To further verify the effectiveness of the improvement, the GCG-GMDH algorithm is applied to the modeling and forecast of other factors in coal price system: coal supply and demand. The results show the GCG-GMDH algorithm can effectively raise the extrapolating performance of the model.