您好,欢迎访问三七文档
当前位置:首页 > 商业/管理/HR > 质量控制/管理 > 统计学第十二章课后作业
第十二章多元线性回归1.利用Excel进行表格计算相关系数设DJIA收益率为x,S&P500收益率为y,将已知表格复制到Excel中,列出计算x2、xy、y2及其合计数的栏目并进行计算,得结果如下:(利用Excel计算进行表格计算的方法类似于标准差的Excel计算)年份DJIA收益率(%)S&P500收益率(%)X2xyy2xy198816.016.6256265.6275.56198931.731.51004.89998.55992.251990-0.4-3.20.161.2810.24199123.930.0571.2171790019927.47.654.7656.2457.76199316.810.1282.24169.68102.0119944.91.324.016.371.69199536.437.61324.961368.641413.76199628.623.0817.96657.8529199724.933.4620.01831.661115.56合计190.2187.94956.25072.825397.83代入相关系数计算公式得:r=2222))nxyxynxxnyy=105072.82190.2187.9104956.2105397.83=0.9481382.解:自变量3个,观察值15个。回归方程:ˆy=657.0534+5.710311X1-0.416917X2-3.471481X3拟合优度:判定系数R2=0.70965,调整的2aR=0.630463,说明三个自变量对因变量的影响的比例占到63%。估计的标准误差yxS=109.429596,说明随即变动程度为109.429596回归方程的检验:F检验的P=0.002724,在显著性为5%的情况下,整个回归方程线性关系显著。回归系数的检验:1的t检验的P=0.008655,在显著性为5%的情况下,y与X1线性关系显著。2的t检验的P=0.222174,在显著性为5%的情况下,y与X2线性关系不显著。3的t检验的P=0.034870,在显著性为5%的情况下,y与X3线性关系显著。因此,可以考虑采用逐步回归去除X2,从新构建线性回归模型。3.(1)回归方程的显著性检验:假设:H0:1=2=0H1:1,2不全等于0SSE=SST-SSR=6724.125-6216.375=507.75F=1SSRpSSEnp=6724.1252507.751021=42.852,7F=4.74,F2,7F,认为线性关系显著。(2)回归系数的显著性检验:假设:H0:1=0H1:1≠0t=11S=2.010.0813=24.7221tnp=2.36,t27t,认为y与x1线性关系显著。(3)回归系数的显著性检验:假设:H0:2=0H1:2≠0t=22S=4.740.0567=83.621tnp=2.36,t27t,认为y与x2线性关系显著。4.1)回归方程为:ˆ88.64+1.6yx(2)回归方程为:12ˆ83.232.291.3yxx(3)不相同,(1)中表明电视广告费用增加1万元,月销售额增加1.6万元;(2)中表明,在报纸广告费用不变的情况下,电视广告费用增加1万元,月销售额增加2.29万元。(4)判定系数R2=0.919,调整的2aR=0.8866,比例为88.66%。(5)回归系数的显著性检验:Coefficients标准误差tStatP-valueLower95%Upper95%下限95.0%上限95.0%Intercept83.230091.57386952.882484.57E-0879.1843387.2758579.1843387.27585电视广告费用工:x1(万元)2.2901840.3040657.5318990.0006531.5085613.0718061.5085613.071806报纸广告费用x2(万元)1.3009890.3207024.0566970.0097610.4765992.1253790.4765992.125379假设:H0:1=0H1:1≠0t=11S=2.290.304=7.530.0255t=2.57,t0.0255t,认为y与x1线性关系显著。(3)回归系数的显著性检验:假设:H0:2=0H1:2≠0t=22S=1.30.32=4.050.0255t=2.57,t0.0255t,认为y与x2线性关系显著。5.1)回归方程为:12ˆ-0.59122.386327.672yxx(2)在温度不变的情况下,降雨量每增加1mm,收获量增加22.386kg/hm2,在降雨量不变的情况下,降雨量每增加1度,收获量增加327.672kg/hm2。(3)1x与2x的相关系数12xxr=0.965,存在多重共线性。9.(1)y与x1的相关系数=0.309,y与x2之间的相关系数=0.0012。对相关性进行检验:相关性销售价格购进价格销售费用销售价格Pearson相关性10.3090.001显著性(双侧)0.2630.997N151515购进价格Pearson相关性0.3091-.853(**)显著性(双侧)0.2630.000N151515销售费用Pearson相关性0.001-.853(**)1显著性(双侧)0.9970.000N151515**.在.01水平(双侧)上显著相关。可以看到,两个相关系数的P值都比较的,总体上线性关系也不现状,因此没有明显的线性相关关系。(2)意义不大。(3)回归统计MultipleR0.593684RSquare0.35246AdjustedRSquare0.244537标准误差69.75121观测值15方差分析dfSSMSFSignificanceF回归分析231778.153915889.083.2658420.073722残差1258382.77944865.232总计1490160.9333Coefficients标准误差tStatP-valueLower95%Upper95%下限95.0%上限95.0%Intercept375.6018339.4105621.106630.290145-363.911115.114-363.911115.114购进价格x10.5378410.210446742.5557110.02520.0793170.9963650.0793170.996365销售费用x21.4571940.667706592.1823860.0496810.0023862.9120010.0023862.912001从检验结果看,整个方程在5%下,不显著;而回归系数在5%下,均显著,说明回归方程没有多大意义,并且自变量间存在线性相关关系。(4)从R2看,调整后的R2=24.4%,说明自变量对因变量影响不大,反映情况基本一致。(5)方程不显著,而回归系数显著,说明可能存在多重共线性。(6)存在多重共线性,模型不适宜采用线性模型。11.dfSSMSFSignificanceF回归分析1187.2519187.251920.22290.000601残差13120.37219.259396总计14307.624Coefficients标准误差tStatP-valueLower95%Upper95%下限95.0%上限95.0%Intercept4.5428571.1501183.9499060.0016622.0581797.0275352.0581797.027535x17.0821431.5748644.4969880.0006013.67985710.484433.67985710.48443(1)回归方程为:ˆ4.547.08yx(2)非易碎品的平均运费为4.54元,易碎品的平均运费为11.62元,易碎品与非易碎品的平均运费差为7.08元。(3)回归方程的显著性检验:假设:H0:1=0H1:1不等于0SSR=187.25195,SSE=120.3721,F=1SSRpSSEnp=6724.1251507.751511=20.22P=0.0006010.05,或者0.051,13F=4.67,F0.051,13F,认为线性关系显著。或者,回归系数的显著性检验:假设:H0:1=0H1:1≠0t=11S=7.081.57=4.5P=0.0006010.05,或者21tnp=0.02513t=2.16,t0.02513t,认为y与x线性关系显著。12.回归统计MultipleR0.943391RSquare0.889987AdjustedRSquare0.871652标准误差96.79158观测值15方差分析dfSSMSFSignificanceF回归分析2909488.4454744.248.539141.77E-06残差12112423.39368.61总计141021912Coefficients标准误差tStatP-valueLower95%Upper95%下限95.0%上限95.0%Intercept732.0606235.58443.1074250.009064218.76641245.355218.76641245.355工龄x1111.220272.083421.5429370.148796-45.8361268.2765-45.8361268.2765性别(1=男,0=女)x2458.684153.45858.580191.82E-06342.208575.1601342.208575.1601拟合优度良好,方程线性显著,工龄线性不显著,性别线性显著。
本文标题:统计学第十二章课后作业
链接地址:https://www.777doc.com/doc-2138698 .html