文章导读
总览 评价 马英钧 1, , 陈楚 2, , 陈岩 2, , 赵东方 2,* ( 1、 华中师范大学数学与统计学学院,武汉 430079; 2、 华中师范大学数学与统计学学院,武汉(430079); ) 摘要: 由于采用累积多元Logistic回归建立的武汉降雨量与太阳黑子的关系模型,
马英钧1,, 陈楚2,, 陈岩2,, 赵东方2,*
(
1、华中师范大学数学与统计学学院,武汉 430079; 2、华中师范大学数学与统计学学院,武汉(430079); )
摘要:
由于采用累积多元Logistic回归建立的武汉降雨量与太阳黑子的关系模型,得到的效果并不理想。本文在分步Logistic回归-"化多元为二元"思想的基础上,提出了基于回归诊断最优的临界值选取和分类次序选取的标准,对传统的多分类Logistic模型进行改进,并给出建立分布Logistic回归模型的一般过程。在此基础上首先利用1951-2010年的降雨量等级和不同期的太阳黑子数建立分步Logistic回归模型。其次利用2005-2010年的实际降雨量检验模型的拟合效果,进而验证了模型的精度较高。然后分析了不同期的太阳黑子数对降雨等级的影响。最后,预测2011-2015年武汉市的降雨量。
关键词:
分步Logistic回归;回归诊断;临界值;拟合检验;Matlab程序
Ma Yingjun1,, Chen Chu2,, Chen Yan2,, Zhao Dongfang2,*
(
1、School of Mathematics and Statistics,Huazhong Normal University, WuHan 430079; 2、School of Mathematics and Statistics,Huazhong Normal University,Wuhan(430079); )
Abstract:
The relational model of the Wuhan rainfall and sunspots established with the cumulative Logistic regression turns out to be unsatisfactory. Based on the substep Logistic regression idea - "multiple regression into two variable regression",this paper suggests taking the optimal regression diagnosis criteria to select the critical values and the orders of multiple classes.It is an improvement on the basis of the traditional multiple regression. And the general processes to establish substep Logistic regression model are given.Firstly,build this substep Logistic regression model to search the relationship between the levels of 1951-2010 Wuhan rainfall and sunplots.Secondly,verify the accuracy of the model to be high by applying the fitting effect of actual rainfall test model of 2005-2010 years. Then analyze the influence of sunspots with different periods on rainfall levels;Finally, forecast Wuhan rainfall bewteen 2011and 2015.
Tag:
点此返回栏目查看更多>>>参考论文