《(完整word版)应用回归分析第四版课后习题答案_全_何晓群_刘文卿.pdf》由会员分享，可在线阅读，更多相关《(完整word版)应用回归分析第四版课后习题答案_全_何晓群_刘文卿.pdf（9页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、实用回归分析第四版第一章回归分析概述1.3 回归模型中随机误差项 的意义是什么？答：为随机误差项，正是由于随机误差项的引入，才将变量间的关系描述为一个随机方程，使得我们可以借助随机数学方法研究y 与 x1,x2.xp的关系，由于客观经济现象是错综复杂的，一种经济现象很难用有限个因素来准确说明，随机误差项可以概括表示由于人们的认识以及其他客观原因的局限而没有考虑的种种偶然因素。1.4 线性回归模型的基本假设是什么？答：线性回归模型的基本假设有：1.解释变量 x1.x2.xp 是非随机的，观测值xi1.xi2.xip是常数。2.等方差及不相关的假定条件为E(i)=0 i=1,2.Cov(i,j)=
2、2 3.正态分布的假定条件为相互独立。4.样本容量的个数要多于解释变量的个数，即 np.第二章一元线性回归分析思考与练习参考答案2.1一元线性回归有哪些基本假定?答：假设 1、解释变量 X 是确定性变量，Y 是随机变量；假设 2、随机误差项 具有零均值、同方差和不序列相关性：E(i)=0 i=1,2,nVar(i)=2i=1,2,nCov(i,j)=0 i j i,j=1,2,n假设 3、随机误差项 与解释变量 X 之间不相关：Cov(Xi,i)=0 i=1,2,n假设 4、服从零均值、同方差、零协方差的正态分布iN(0,2)i=1,2,n2.3 证明（2.27式），ei=0，eiXi=0。证
3、明：niiiniXYYYQ121021)?()?(其中：即：ei=0，eiXi=02.5 证明0?是 0的无偏估计。证明：)1)?()?(1110niixxiniiYLXXXYnEXYEE)(1()1(1011iixxiniixxiniXLXXXnEYLXXXnE01010)()1()1(ixxiniixxiniELXXXnLXXXnE2.6 证明证明：)()1()1()?(102110iixxiniixxiniXVarLXXXnYLXXXnVarVar2222121)(2)1(xxxxixxiniLXnLXXXnLXXXn2.7 证明平方和分解公式：SST=SSE+SSR证明：2.8 验证三
4、种检验的关系，即验证：（1）21)2(rrnt；（2）2221?)2/(1/tLnSSESSRFxx01?iiiiiYXeYY)1()1()?(2221220 xxniiLXnXXXnVarniiiiniiYYYYYYSST1212?()?niiiniiiiniiYYYYYYYY12112)?)(?2?SSESSR)Y?YYY?n1i2iin1i2i0100?QQ文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6
5、ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6
6、 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X
7、6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5
8、X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A
9、5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1
10、A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y
11、1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6证明：（1）22?22?(2)(2)?1yyxxyyxxxxxxrLLrLLnrnrtSSE LnSSE nSSE SSTLr（2）22222011111111?()()()()nnnniiiixxiiiiSSRyyxyyxxyxxL2212?/1?/(2)xxLSSRFtSSE n2.9 验证（2.63
12、）式：2211)L)xx(n()e(Varxxii证明：0112222222?var()var()var()var()2cov(,)?var()var()2cov(,()()()112()11iiiiiiiiiiiiixxxxixxeyyyyyyyxy yxxxxxxnLnLxxnL其中：222221111)(1()(1)(,()()1,()(?,(),()(?,(xxixxiniixxiiiniiiiiiiiLxxnLxxnyLxxyCovxxynyCovxxyCovyyCovxxyyCov2.10 用第 9 题证明是2的无偏估计量证明：2221122112211?()()()22()111
13、var()1221(2)2nniiiinniiiixxEE yyE ennxxennnLnn第三章2?22nei文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X
14、6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5
15、X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A
16、5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1
17、A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y
18、1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6
19、Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F
20、6Y1A5X6 ZC10U8T9A1Y61.一个回归方程的复相关系数R=0.99，样本决定系数 R2=0.9801，我们能判断这个回归方程就很理想吗？答：不能断定这个回归方程理想。因为：1.在样本容量较少，变量个数较大时，决定系数的值容易接近1，而此时可能 F 检验或者关于回归系数的t 检验，所建立的回归方程都没能通过。2.样本决定系数和复相关系数接近于1 只能说明Y 与自变量X1,X2,Xp 整体上的线性关系成立，而不能判断回归方程和每个自变量是显著的，还需进行F检验和 t 检验。3.在应用过程中发现，在样本容量一定的情况下，如果在模型中增加解释变量必定使得自由度减少，使得 R2往往增大，因
21、此增加解释变量（尤其是不显著的解释变量）个数引起的R2的增大与拟合好坏无关。2.被解释变量Y的期望值与解释变量kXXX,21的线性方程为：01122()kkE YXXX(3-2)称为多元总体线性回归方程，简称总体回归方程。对于n组观测值),2,1(,21niXXXYkiiii，其方程组形式为：01122,(1,2,)iiikkiiYXXXin21?*,1,2,.,)jjyynjjjiLjpLLXjjij其中:(X文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:C
22、D7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:
23、CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码
24、:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编
25、码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档
26、编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文
27、档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6
28、文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6(3-3)即nknknnnkkkkXXXYXXXYXXXY2211022222121021121211101其矩阵形式为nYYY21=knnnkkXXXXXXXXX212221212111111k210+n21即YX(3-4)其中1nYnYYY21为被解释
29、变量的观测值向量；)1(knXknnnkkXXXXXXXXX212221212111111为解释变量的观测值矩阵；(1)1kk210为总体回归参数向量；1nn21为随机误差项向量。多元回归线性模型基本假定：课本P57 第四章4.3 简述用加权最小二乘法消除一元线性回归中异方差性的思想与方法。答：普通最小二乘估计就是寻找参数的估计值使离差平方和达极小。其中每个平方项的权数相同，是普通最小二乘回归参数估计方法。在误差项等方差不相关的条件下，普通最小二乘估计是回归参数的最小方差线性无偏估计。然而在异方差的条件下，平方和中的每一项的地位是不相同的，误差项的方差大的项，在残差平方和中的取值就偏大，作用就
30、大，因而普通最小二乘估计的回归线就被拉向方文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU1
31、0F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU
32、10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 H
33、U10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4
34、HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4
35、 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S
36、4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6差大的项，方差大的项的拟合程度
37、就好，而方差小的项的拟合程度就差。由 OLS求出的仍然是的无偏估计，但不再是最小方差线性无偏估计。所以就是：对较大的残差平方赋予较小的权数，对较小的残差平方赋予较大的权数。这样对残差所提供信息的重要程度作一番校正，以提高参数估计的精度。加权最小二乘法的方法：4.4 简述用加权最小二乘法消除多元线性回归中异方差性的思想与方法。答：运用加权最小二乘法消除多元线性回归中异方差性的思想与一元线性回归的类似。多元线性回归加权最小二乘法是在平方和中加入一个适当的权数iw，以调整各项在平方和中的作用，加权最小二乘的离差平方和为：niippiiipwxxywQ1211010)(),(（2）加权最小二乘估计就是
38、寻找参数p,10的估计值pwww?,?,?10使式（2）的离差平方和wQ达极小。所得加权最小二乘经验回归方程记做ppwwwwxxy?110（3）220111?()()NNwiiiiiiiiQwyywyx22_1_2_02222()()?()?1111,iiNwiiiwiwiwwwwwkxiiiimiiimiw xxyyxxyxwkxxkxwx1Ni=11表示=或文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 Z
39、C10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6
40、ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6
41、 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X
42、6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5
43、X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A
44、5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1
45、A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6多元回归模型 加权最小二乘法的方法:首先找到权数iw，理论上最优的权数iw为误差项方差2i的倒数,即21iiw（4）误差项方差大的项接受小的权数，以降低其在式（2）平方和中的作用;误差项方差小的项接受大的权数，以提高其在平方和中的作用。由（2）式求出的加权最小二乘估计pwww?,?,?10就是参数p,10的
46、最小方差线性无偏估计。一个需要解决的问题是误差项的方差2i是未知的,因此无法真正按照式（4）选取权数。在实际问题中误差项方差2i通常与自变量的水平有关(如误差项方差2i随着自变量的增大而增大),可以利用这种关系确定权数。例如2i与第 j 个自变量取值的平方成比例时,即2i=k2ijx 时,这时取权数为21ijixw（5）更一般的情况是误差项方差2i与某个自变量jx(与|ei|的等级相关系数最大的自变量)取值的幂函数mijx 成比例，即2i=kmijx,其中 m 是待定的未知参数。此时权数为mijixw1（6）这时确定权数iw的问题转化为确定幂参数m的问题，可以借助 SPSS软件解决。第五章5.
47、3 如果所建模型主要用于预测，应该用哪个准则来衡量回归方程的优劣？答：如果所建模型主要用于预测，则应使用pC统计量达到最小的准则来衡量回归方程的优劣。文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7
48、D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H
49、7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7
50、H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD7H7D6S10S4 HU10F6Y1A5X6 ZC10U8T9A1Y6文档编码:CD