Combined Forecasting Model of Soybean Seed Average Diameter Based on the Shapley Value
Dun Guo-qian · Dadou kexue · 2014
Average diameter of soybean seed,as an important reference of breeding seed plate designing,the precise construction of average diameter of soybean seed forecasting model is very important. The linear regression model,grain weight physical model,quadratic regression model were built to predict the average diameter of soybean seed,and then contrastive analyzed prediction error of every forecasting model. Then the weight of every single forecasting model was determined by using Shapley value method,the combination forecasting model was built and used to predict average diameter of 5 Dongnong-52 soybean seed samples which were selected randomly. The results indicated that combination forecasting model's prediction accuracy was higher than the selected individual forecasting models,the combined model had a smaller average error and decreased the fluctuation of error,the prediction of average diameter of soybean seed was feasible and effective. Meanwhile,the research provided a practical new method for relative physical parameters' prediction of other crop seeds.