资源描述
回归分析作业
一、利用软件计算
1、
数据文件“资产评估1”提供了35家上市公司资产评估增值的数据。
num---公司序号
pg---- 资产评估增值率
gz----固定资产在总资产中所占比例
fz----权益与负债比
bc----总资产投资报酬率
gm---公司资产规模(亿元)
a. 建立关于资产评估增值率的四元线性回归方程,并通过统计分析、检验说明所得方程的有效性,解释各回归系数的经济含义。
b. 剔除gz变量,建立关于资产评估增值率的三元线性回归方程,与a中的模型相比较,那个更为实用有效,说明理由。
解:
a. SPSS数据如下
Model Summary
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.871a
.759
.727
.0787500
a. Predictors: (Constant), 公司规模, 权益与负债比, 固定资产比重, 总资产投资报酬率
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
.586
4
.146
23.609
.000a
Residual
.186
30
.006
Total
.772
34
a. Predictors: (Constant), 公司规模, 权益与负债比, 固定资产比重, 总资产投资报酬率
b. Dependent Variable: 资产评估增值率
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
B
Std. Error
Beta
1
(Constant)
.396
.145
2.736
.010
固定资产比重
.079
.082
.092
.972
.339
权益与负债比
.062
.016
.416
3.918
.000
总资产投资报酬率
.602
.130
.493
4.618
.000
公司规模
-.044
.014
-.304
-3.201
.003
a. Dependent Variable: 资产评估增值率
Residuals Statisticsa
Minimum
Maximum
Mean
Std. Deviation
N
Predicted Value
-.084652
.494055
.172240
.1312429
35
Residual
-1.5000236E-1
.1493797
.0000000
.0739727
35
Std. Predicted Value
-1.957
2.452
.000
1.000
35
Std. Residual
-1.905
1.897
.000
.939
35
a. Dependent Variable: 资产评估增值率
由Model Summary和ANOVA表 可知,R为0.871,决定系数R2为0.759,校正决定系数为0.727。拟合的回归模型F值为23.609,P值为0,所以拟合的模型是有统计意义的。
从系数的t检验可以看出,只有固定资产比重的sig值=0.339>0.05,说明只有固定资产比重对资产评估增值率的影响是不显著的,其他自变量对固定资产增值的比率均有显著的影响。
线性回归方程为:
pg=0.396+0.079gz+0.063fz+0.602bc-0.044gm
α1=0.079表示,在权益与负债比、总资产投资报酬率和公司规模不变的条件下,固定资产比重每增加1个单位,资产评估增值率增加0.079。
α2=0.063表示,在固定资产比重、总资产投资报酬率和公司规模不变的条件下,权益与负债比每增加1个单位,资产评估增值率增加0.063。
α3=0.602表示,在固定资产比重、权益与负债比和公司规模不变的条件下,总资产投资报酬率每增加1个单位,资产评估增值率增加0.602。
α4=-0.044表示,在固定资产比重、权益与负债比和总资产投资报酬率不变的条件下,公司规模每增加1亿元,资产评估增值率减少0.044
b.
Model Summaryb
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.867a
.751
.727
.0786809
a. Predictors: (Constant), 公司规模, 权益与负债比, 总资产投资报酬率
b. Dependent Variable: 资产评估增值率
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
.580
3
.193
31.218
.000a
Residual
.192
31
.006
Total
.772
34
a. Predictors: (Constant), 公司规模, 权益与负债比, 总资产投资报酬率
b. Dependent Variable: 资产评估增值率
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
B
Std. Error
Beta
1
(Constant)
.376
.143
2.628
.013
权益与负债比
.063
.016
.422
3.981
.000
总资产投资报酬率
.600
.130
.491
4.607
.000
公司规模
-.040
.013
-.275
-3.052
.005
a. Dependent Variable: 资产评估增值率
Residuals Statisticsa
Minimum
Maximum
Mean
Std. Deviation
N
Predicted Value
-.062589
.511513
.172240
.1305841
35
Residual
-1.6124609E-1
.1453045
.0000000
.0751295
35
Std. Predicted Value
-1.798
2.598
.000
1.000
35
Std. Residual
-2.049
1.847
.000
.955
35
a. Dependent Variable: 资产评估增值率
由Model Summary和ANOVA表 可知,相关系数R为0.867,决定系数R2为0.751,校正决定系数为0.727。
从系数的t检验可以看出,该模型的回归系数都通过检验。所以,剔除 gz 变量,建立关于资产评估增值率的三元线性回归方程为:
pg=0.376+0.063fz+0.600bc-0.040gm
b更为有效实用,因为所有的回归系数都通过了t检验,并且b模型估计的标准误较小。
2、
数据文件“房产销售”提供了20件房地产的销售价格和评估的数据(美元):
y----销售价格; x1----地产评估价值; x2----房产评估价值;x3----面积(平方英尺)。
a. 建立适当的关于销售价格的多元线性回归模型.
b. 利用模型预测地产评估价值为2000,房产评估价值为12000,面积为1100的销售价格,并给出预测值的95%的置信区间。
c. 通过对模型的统计检验说明预测值的可信度。
解:
a. SPSS数据如下
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
B
Std. Error
Beta
1
(Constant)
1475.648
5742.859
.257
.800
地产价值
.814
.512
.193
1.589
.132
房产价值
.821
.211
.557
3.890
.001
面积
13.509
6.583
.277
2.052
.057
a. Dependent Variable: 销售价格
由图表所知,地产价值的sig值过高,所以地产价值对销售价格的影响不显著。
把地产价值剔除后,所得的数据如下:
Model Summaryb
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.939a
.881
.867
8262.430
a. Predictors: (Constant), 面积, 房产价值
b. Dependent Variable: 销售价格
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
8.623E9
2
4.311E9
63.153
.000a
Residual
1.161E9
17
6.827E7
Total
9.783E9
19
a. Predictors: (Constant), 面积, 房产价值
b. Dependent Variable: 销售价格
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
105.382
5927.158
.018
.986
-12399.827
12610.592
房产价值
.961
.200
.651
4.797
.000
.538
1.384
面积
16.348
6.615
.336
2.472
.024
2.392
30.304
a. Dependent Variable: 销售价格
Residuals Statisticsa
Minimum
Maximum
Mean
Std. Deviation
N
Predicted Value
21387.19
1.06E5
5.67E4
21303.096
20
Residual
-1.582E4
1.387E4
.000
7815.476
20
Std. Predicted Value
-1.656
2.330
.000
1.000
20
Std. Residual
-1.915
1.679
.000
.946
20
a. Dependent Variable: 销售价格
由Coefficients表所知,回归方程为:y=105.382+0.961x2+16.348x3
b.
解:通常先做enter,然后做逐步
(1)对原数据进行回归分析,得到回归方程为:y=105.382+0.961x2+16.348x3
(2)地产评估价值为2000,房产评估价值为12000,面积为1100的销售价格的95%的置信区间为:(21468.99197,37776.93332)。
(3)该模型的Adjusted R Square=0.867,也就是这两个自变量可以解释86.7%的因变量变差,应该说是预测的可信度比较高;并且残差符合正态性、独立性和方差齐次性,模型成立,即有95%的可能性b的预测值在区间21468.99197-37776.93332内。
3、
大多数公司都提供了β估计值,以反映证券的系统风险。一种股票的β值所测量的是这种股票的回报率与整个市场平均回报率之间的关系。这个指标的名称就来自简单线性回归中的斜率参数β。在这种回归中,因变量是股票回报率(Y)。而自变量则是市场回报率(X)。 值大于1的股票被称为“攻击性”证券,因为它们的回报率变动(向上或向下)得比整个市场的回报率快。相反,β值小于1的股票被称为“防御性”证券,因为它们的回报率变动的比市场回报率慢。 值接近1的股票被称为“中性”证券,因为它们的回报率反映市场回报率。下面表中的数据是随机抽选的7个月内某只特定的股票的月回报率及整个市场的回报率。试对这些数据完成简单线性回归分析。根据你的分析结果,你认为这只股票是属于攻击性,防御性,还是中性的股票?
月 股票回报率Y 市场回报率X
1 12.0 7.2
2 –1.3 0.0
3 2.5 2.1
4 18.6 11.9
5 9.0 5.3
6 –3.8 –1.2
7 –10.0 –4.7
解:
SPSS数据如下:
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
B
Std. Error
Beta
1
(Constant)
-1.329
.323
-4.109
.009
X
1.762
.054
.998
32.539
.000
a. Dependent Variable: Y
得到回归方程:y=1.762x-1.329。
β值为线性回归斜率参数1.762>1,所以,该股票属于“攻击性股票”。
4、
参考上题。股票的β值是否依赖于计算回报率的时间长度?因为有些经济商号用的是按月数据计算的β值,另一些经济商号则用按年数据计算的β值,所以这个问题对投资者来说很重要。H.莱维分别研究了三类股票的时间长度(月)和平均β值。将时间长度从一个月逐步增加到30个月,莱维计算了1946---1975年间144只股票的回报率。根据他所得的β值,这144只股票中有38只攻击性股票,38只防御性股票,以及68只中性股票。下表中给出的这三类股票对不同时间水平的平均β值。
A、 对于攻击性股票、防御性股票和中性股票三种情况,分别求表达平均β值Y与时间长度X之间关系的最小二乘简单线性回归方程。
B、 对每一类股票检验假设:时间长度是平均β值的有效线性预测器,检验时用α=0.05。
C、 对每一类股票,构造直线斜率的95%置信区间,哪只股票的β值随时间长度的增大而线性增大?
时间长度X,月
β值, Y
攻击性股票 防御性股票 中性股票
1
3
6
9
12
15
18
24
30
1.37 0.50 0.98
1.42 0.44 0.95
1.53 0.41 0.94
1.69 0.39 1.00
1.83 0.40 0.98
1.67 0.38 1.00
1.78 0.39 1.02
1.86 0.35 1.14
1.83 0.33 1.22
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
1.451
.059
24.392
.000
1.310
1.591
x
.016
.004
.856
4.377
.003
.007
.025
a. Dependent Variable: y
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
.459
.013
34.178
.000
.428
.491
x
-.005
.001
-.901
-5.488
.001
-.007
-.003
a. Dependent Variable: y
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
.911
.025
37.083
.000
.853
.969
x
.009
.002
.906
5.672
.001
.005
.012
a. Dependent Variable: y
5
个人计算机(PC机)正以非凡的技术在发展,PC机的零售价格也是这样。由于购买时间和机器特点不同,一台PC机的零售价格可能发生戏剧性的变化。不久前收集了一批IBM PC机和IBM PC兼容机的零售价格数据,共有N=60,见数据文件“计算机价格”。这些数据被用来拟合多元回归
E(Y)=β0+β1 x1+β2x2
其中:Y=零售价格(美元)
x1=微处理器速度(兆赫)
a、 试写出最小二乘预测方程。
b、 此模型是否适合于预测?用α=0.10进行检验。
c、 构造β1 的90%置信区间,并对此区间作出解释。
d、 本模型中的CPU芯片(x2)是否是价格(Y)的有效预测器?用α=0.10进行预测。
Model Summaryb
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.610a
.373
.350
962.967
a. Predictors: (Constant), 芯片, 速度
b. Dependent Variable: 价格y
Model Summaryb
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.610a
.373
.350
962.967
a. Predictors: (Constant), 芯片, 速度
b. Dependent Variable: 价格y
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
-68.509
1461.468
-.047
.963
-2996.181
2859.162
速度
108.237
21.198
.582
5.106
.000
65.772
150.702
芯片
2.486
4.174
.068
.596
.554
-5.876
10.848
a. Dependent Variable: 价格y
6、
在工厂中,准确估计完成一项作业所需的工时数对于诸如决定雇佣工人的数量,确定向客户报价的最后期限,或者作出与预算有关的成本分析决策等决策管理来说是极端重要的。一名锅炉筒制造商想预测在一些在未来预测项目中装配锅炉筒所需的工时数。为了用回归方法实现此目标,他收集了35个锅炉的项目数据(数据文件“锅炉”)。除工时(Y)外,被测量的变量有锅炉工作容量(X1=磅/小时),锅炉设计压力(X2=磅/平方英寸),锅炉的类型(X3=1,如在生产领域装配;X3=0,如在使用领域装配),以及炉筒类型(X4=1,蒸汽炉筒;X4=0,液体炉筒)。
A、 试检验假设:锅炉容量(X1)与工时数(Y)之间有正线性关系。
B、 试检验假设:锅炉压力(X3)与工时数(Y)之间有正线性关系。
C、 构造β1的95%置信区间并对结果做出解释。
D、 构造β3的95%置信区间。
Correlations
工时y
容量x1
工时y
Pearson Correlation
1
.827**
Sig. (2-tailed)
.000
N
35
35
容量x1
Pearson Correlation
.827**
1
Sig. (2-tailed)
.000
N
35
35
**. Correlation is significant at the 0.01 level (2-tailed).
Correlations
工时y
压力x2
工时y
Pearson Correlation
1
.657**
Sig. (2-tailed)
.000
N
35
35
压力x2
Pearson Correlation
.657**
1
Sig. (2-tailed)
.000
N
35
35
**. Correlation is significant at the 0.01 level (2-tailed).
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
-3727.268
1227.784
-3.036
.005
-6234.737
-1219.800
容量x1
.009
.001
.903
9.491
.000
.007
.011
压力x2
1.898
.661
.388
2.873
.007
.549
3.247
炉类型x3
3410.104
926.871
.531
3.679
.001
1517.180
5303.027
筒类型x4
2118.726
314.805
.392
6.730
.000
1475.809
2761.644
a. Dependent Variable: 工时y
7
Cushman & Wakefield 股份有限公司,采集了美国市场上办公用房的空闲率和租金率的数据。对于18个选取的销售地区,这些地区的中心商业区的综合空闲率(%)和平均租金率(美元/平方英尺)的数据(The Wall Journal Almanac1988)见文件“办公用房”。
a. 用水平轴表示空闲率,对这些数据画出散点图。
b. 这两个变量之间显出什么关系吗?
c. 求出在办公用房的综合空闲率已知时,能用来预测平均租金率的估计的回归方程。
d. 在0.05显著水平下检验关系的显著性。
e. 估计的回归方程对数据的拟合好吗?请解释。
f. 在一个综合空闲率是25%的中心商业区,预测该市场的期望租金率。
g. 在劳德代尔堡的中心商业区,综合空闲率是11.3%,预测劳德代尔堡的期望租金率。
Correlations
综合空闲
平均租金
综合空闲
Pearson Correlation
1
-.659**
Sig. (2-tailed)
.003
N
18
18
平均租金
Pearson Correlation
-.659**
1
Sig. (2-tailed)
.003
N
18
18
**. Correlation is significant at the 0.01 level (2-tailed).
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
37.075
3.528
10.510
.000
29.596
44.553
综合空闲
-.779
.222
-.659
-3.504
.003
-1.251
-.308
a. Dependent Variable: 平均租金
Variables Entered/Removedb
Model
Variables Entered
Variables Removed
Method
1
综合空闲a
.
Enter
a. All requested variables entered.
b. Dependent Variable: 平均租金
Model Summaryb
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
Durbin-Watson
1
.659a
.434
.399
4.88474
2.456
a. Predictors: (Constant), 综合空闲
b. Dependent Variable: 平均租金
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
292.914
1
292.914
12.276
.003a
Residual
381.771
16
23.861
Total
674.685
17
a. Predictors: (Constant), 综合空闲
b. Dependent Variable: 平均租金
Residuals Statisticsa
Minimum
Maximum
Mean
Std. Deviation
N
Predicted Value
19.3099
32.3998
25.3917
4.15093
18
Std. Predicted Value
-1.465
1.688
.000
1.000
18
Standard Error of Predicted Value
1.157
2.308
1.590
.361
18
Adjusted Predicted Value
19.2299
32.0616
25.4049
4.20352
18
Residual
-6.36692
8.37020
.00000
4.73890
18
Std. Residual
-1.303
1.714
.000
.970
18
Stud. Residual
-1.342
1.798
-.001
1.016
18
Deleted Residual
-6.74506
9.21847
-.01321
5.19557
18
Stud. Deleted Residual
-1.379
1.949
.013
1.052
18
Mahal. Distance
.009
2.850
.944
.873
18
Cook's Distance
.000
.164
.047
.048
18
Centered Leverage Value
.001
.168
.056
.051
18
a. Dependent Variable: 平均租金
8.
PJH&D公司正在决定是否为公司新的文字处理系统签订一项维修合同。公司的管理人员认为,维修费用与该系统的使用时间有关。采集的每周时间(小时)和面维修费用(千美元)的统计资料见“文字处理系统”。
a. 求出年维修费用对于每周使用时间的估计的回归方程。
b. 在0.05显著水平下, 检验在(a)中求出关系的显著性。
c. PJH&D公司预期每周使用文字处理系统的时间是30小时,求出该公司的年维修费用的 95%的预测区间。
d. 如果维修合同 的费用是每年3000美元,你建议签订这个合同吗,为什么?
Model Summary
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.925a
.856
.839
4.2102
a. Predictors: (Constant), 使用时间
e.
f.
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
846.092
1
846.092
47.732
.000a
Residual
141.808
8
17.726
Total
987.900
9
a. Predictors: (Constant), 使用时间
b. Dependent Variable: 年维修费
g.
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
B
Std. Error
Beta
1
(Constant)
10.675
3.710
2.877
.021
使用时间
.946
.137
.925
6.909
.000
a. Dependent Variable: 年维修费
9.
对于一个较大的人口密集的地区 ,当地交通部门想要确定公共汽车的使用时间和年维修费用之间是否存在某种关系。由10辆公共汽车组成一个样本,采集的数据见文件“交通”。
a. 利用最小二乘法求出估计的回归方程。
b. 在=0.05的显著水平下, 通过检验是否看出二变量之间存在一个显著的关系。
c. 最小二乘法回归线给出了观测数据一个好的拟合吗?请做出解释。
d. 如果有一辆特定的公共汽车已使用了4年,求出这辆车年维修费用的一个95%的预测区间。
Model Summary
Model
R
R Square
Adjusted R Square
Std. Error of the Estimate
1
.934a
.873
.857
75.498
a. Predictors: (Constant), 使用时间
ANOVAb
Model
Sum of Squares
df
Mean Square
F
Sig.
1
Regression
312050.000
1
312050.000
54.746
.000a
Residual
45600.000
8
5700.000
Total
357650.000
9
a. Predictors: (Constant), 使用时间
b. Dependent Variable: 年维修费
Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t
Sig.
95% Confidence Interval for B
B
Std. Error
Beta
Lower Bound
Upper Bound
1
(Constant)
220.000
58.481
3.762
.006
85.143
354.857
使用时间
131.667
17.795
.934
7.399
.000
90.631
172.702
a. Dependent
展开阅读全文