多元样条逐步回归模型在夏收中期预报中的应用

在夏收期间的中期预报中，使用了非线性多元样条函数回归预报模型。该方法保留了线性统计预报模型中多元分析，逐步筛选因子和显著性检验等特点，又吸取了数值逼近中样条函数处理复杂非线性函数的一些优点。经过与线性回归的对比和１９９８年的预报，在历史拟合率和预报效果上的提高较为明显。

APPLICATION OF MULTIPLE SPLINE PROGRESSIVE REGRESSION MODEL TO MEDIUM - RANGE PREDICTION OF SUMMER HARVEST

The nonlinear multiple spline progressive regression model is used in the medium-range prediction of sununer harvest. It retains multivariate analysis,screening factories and significance test of linear statistical prediction, and absorbs the advantage of splines that can deal with the nonlinear function in numerical approach. The improvement of fitting and results is obvious through the comparative analysis of 1iilear regression and forecasting in 1998.