CAJ | 학술논문

采用一次预报最佳子集回归方法，设计并实施了筛选预报因子的向后法和引进、剔除的双重检验逐步筛选法两种计算方案。试验表明，一次预报最佳子集回归与通常的逐步回归分析相比具有明显的优越性。通过对新安江流域（10站平均）汛期（4～9月）降水量资料的分析，从中提取隐含的周期分量，采用上述方案进行实例试报，结果表明，这一方案具有一定的预测能力和应用前景。
채용일차예보최가자집회귀방법，설계병실시료사선예보인자적향후법화인진、척제적쌍중검험축보사선법량충계산방안。시험표명，일차예보최가자집회귀여통상적축보회귀분석상비구유명현적우월성。통과대신안강류역（10참평균）신기（4～9월）강수량자료적분석，종중제취은함적주기분량，채용상술방안진행실례시보，결과표명，저일방안구유일정적예측능력화응용전경。
Two calculation schemes,backward algorithm screening prediction and double chest stepwise screening for algorithm of introduction and rejection prediction,were designed and applied using the method of best subset regression for single prediction.The study shows that the method has advantages over other stepwise regression analyses.The method was used to analyze the rainfall data of Xinanjiang River drainage basin during the rainy season (April to September).The results indicate that the method has some predictive ability.