统计软件作业

迟恬2009329709国贸2班

第二章描述统计分析与参数假设检验

Exercise2-1

(1)

1、打开源文件Exercise文件中的Exercise2-1,双击inc,找到题中所给数据。

2、点ViewDescriptiveStatisticsHistogramandStat得到直方图

3、单击View选择——DescriptiveStatistics——StatsTable得到统计表

View][Proc][object]Properties][PrintRName[[Freeze]Sample](Genr][sheet](Graph][stats][ldent

INC

Mean31.27800S

Median25.70000

Maximum85.50000

Minimum6.420000

Std.Dev.22.37583

Skewness1.368930

Kurtosis4.071719

Jarque-Bera7.203714

Probability0.027273■

Sum625.5600

SumSq.Dev.9512.881

Observations20

id

(2)

单击ViewDescriptiveStatisticsStatisticByClassification得到分组表格，在分组变量处输入

“edu”得到分组统计描述。

即收入在12-16的教育年限分布人数最多。

□覆盟兴盘归阂fe?盘附

Mew][Prod[objecH|Properties][PrielRNGmeRFreeze][Sample][Genr][sheeH[Graph][^E7|[ldent

DescriptiveStatisticsforINC

CategorizedbyvaluesofEDU

Date:06/16/12Time:11:34

Sample:120

Includedobservations:20

EDUMeanStd.Dev.Obs.

1017.500003.3941132

1219.3371413.444897

1648.0142927.627251

1838.1000013.576452

2021.450005.3033012

All31.2780022.3758320

(3)

单击ViewDiscriptiveStatistics&TestsSimpleHypothesisTest”输入Mean15>Variance81得到

检验结果。

'iew][Proc][objecH[Properties][Print(Name正reeze]【Sample][Genr)|sheeH[Grdphgtats][ldenl:

HypothesisTestingforINC

Date:06/16/12Time:11:44

Sample:120

Includedobservations:20

TestofHypothesis:Mean=15.00000

SampleMean=31.27800

SampleStd.Dev.=22.37583

MethodValueProbability

^statistic3.253395—0.0042

TestofHypothesis:Variance=81.00000

SampleVariance=500.6779

MetiiodValueProbability

VarianceRatio117.4430—0.0000

(4)

1、单击ViewGraphTypeDistributionEmpiricalCDF得到经验累积分布图

2、单击ViewGraphTypeQuantile-Quantile得到序列Q-Q散点图

E］回区

[view][ProW[obj8ct][prop8rgs][PrintJ[NameUFre8zeJ6amp阳][qraph][^i?)Udent]

-20

0102030405060708090

QuantilesofINC

3^单击ViewDiscriptiveStatistics&TestsEmpiricalDistributionTesto由表可以看出其服从正

态分布。

00s

[viewJ[Proc[objecHFroperties||Print]|Nam^Freece][Sample[[GraphgtatsRldent|

EmpiricalDistributionTestforINC

Hypothesis:Normal

Date:06/16/12Time:11:53

Sample:120

Includedobservations:20

MethodValueAdj.ValueProbability

Lilliefors(D)0.246042NA0.0025

Cramer-vonMises(W2)0.1991330.2041110.0048

Watson(U2)0.1684750.1726870.0074

Anderson-Darling(A2)1.2068481.2588930.0028

Method:MaximumLikelihood-d.f.corrected(ExactSolution)

ParameterValueStd.Errorz-StatisticProb.

MU31.278005.0033896.2513630.0000

SIGMA22.375833.6298406.1644140.0000

Loglikelihood-90.03840Meandependentvar.31.27800

No.ofCoefficients2S.D.dependentvar.22.37583

(5)

在Eviews命令窗口中输入命令groupglincedu”按enter生成新的序列组

£il«EditQb”ctVit*trocQaickOfitionx量iad”削

groupg1incedu

O

(6)

1、双击gl打开序列组窗口，单击ViewDescriptiveStatisticscommonsample得到描述性统计

分析。

2、相关系数:

View|Proc|Object|Print|Name|Freeze|5ampte|Sheet|Stats|Spec|

CorrelationMatiix

INCEDU

INC1.0000000.338146

EDU0.3381461.000000

3、协方差:

CovarianceMatiix

INCEDU

INC475.644022.41720

EDU22,417209.240000

(7)

单击ViewGraphTypeScatterRegressionLine得到序列组gl的回归散点图。根据散点

图可以看出序列inc和edu成正相关关系，但并不是高度正相关关系。

Exercise2-2

(1)

1、打开源文件Exercise文件中的Exercise2-2,分别双击ggdp、gcs,找到题中所给数据。

2、分别单击ViewDescriptiveStatisticsHistogramandStat得到两个直方图

□器a?捌颛^^75遹更舒遨就的唠空芍施：!堡一日]0同

[view][Proc]|objecH[PropertiesJ[PrinH|Name帕eeze][Sample][^i7||sheet[[Graph

(2)

1、单击ViewDiscriptiveStatistics&TestsSimpleHypothesisTest分别输入Mean10.2、Mean7.6

得到下表。

Q@0

ViewProcObjectPropertiesPrintNameFruweidrripleGenrSheetGraphStatsIdent

HypothesisTestingforGGDP

Date:06/16/12Time:13:32

Sample:19922000

Includedobservations:9

TestofHypothesis:Mean=10.20000

SampleMean=10.23333

SampleStd.Dev.=2.628212

MethodValueProbability

t-statistic0.038049-0.9706

Meantestassumption

Meantestwilusea

knownstandard

deviationifsupplied.

Enters.d.

fknown:

[邺]|Cancel]

(3)1、关闭窗口，单击ObjectNewObjectGroup并将其命名为gl

2、对序列组的序列进行定义，第一列定义为"ggdp"按enter键，第二列定义为“gcs"按enter键。

□

|view||Proc||object||Print|[Save|[Details+/-1|show][Fetch]|store)|Delete||Genr]|SampleI

Range:19922000--9obsDisplayFilter:*I

Sample:19922000-9obs

固g1|口"如山Gl之工义支片

I£J.'C11jJ

I2gcs_______

0ggdp'「TVHot"jecFrinriif.jamefreeze；uerauirjportIranspose||tdir+/-1|brnpi+1-

0resid[EE

obsGGDPGCS

obsGGDPOCSQI

199214.2000012.90000

199313.500008.100000

199412.600004.300000

199510.500007.500000

19969.6000009.100000

19978.8000004200000

19987.8000005.500000

19997.1000007.900000

20008.0000009.100000

\Untitled/

QI

0

(4)

2、单击View选择TestofEquality选择Variance单击OK。得到方差检验结果。GGDP与GCS的方差

相等。

Mew][Proc][object)[Print][Name[Freeze115amplegheeHBtatsgpec]

TestforEqualityofVariancesBetweenSeries

Date:06/16/12Time:13:39

Sample:19922000

Includedobservations:9

MethoddfValueProbability

F-test(8.8)1.0820770.9139

Siegel-Tukey0.0883480.9296

Bartlett10.0117100.9138

Levene(1.16)0.0763590.7858

Brown-Forsythe(1.16)0.0395820.8448

CategoryStatistics

MeanAbs.MeanAbs.MeanTukey-

VariableCountStd.Dev.MeanDiff.MedianDiff.SiegelRank

GGDP92.6282122.1925932.1222229.666667

GCS92.7339431.9975311.9666679.333333

All182.9279192.0950622.0444449.500000

Bartlettweightedstandarddeviation:2.681599

3、同理，选择Mean得到均值检验结果，均值也相等。

MewllProcgbjecH[Print](Name[[Freeze]15ampIe回eetJBtatsgpec]

TestforEqualityofMeansBetweenSeries

Date:06/16/12Time:13:43

Sample:19922000

Includedobservations:9

MethoddfValueProbability

t-test162.0655600.0555

Satterthwaite-Welcht-test*15.975172.0655600.0555

AnovaF-test(1.16)4.2665380.0555

WelchF-test*(1,15.9752)4.2665380.0555

*Testallowsforunequalcellvariances

AnalysisofVariance

SourceofVariationdfSumofSq.MeanSq.

Between130.6805630.68056

Within16115.05567.190972

Total17145.73618.572712

CategoryStatistics

Std.Err.

VariableCountMeanStd.Dev.ofMean

GGDP-910.233332.6282120.876071

GCS97.6222222.7339430.911314

-Ali-188.9277782.9279190.690117

第三章简单线性回归分析

Exercise3-1

(1)1、打开源文件Exercise文件中的Exercise3-1。

2、点Quick选择EquationEstimation使用列表形式对方程进行设定。输入"peonscpdinc”得到下表。

从回归结果可以看出，自变量pdinc能够解释因变量peons93.6%的变化。Pdinc每增加一个单位，peons

的平均值增加0.75811,回归参数的t检验在统计上是显著的，说明估计的回归方程是正确的。

口白!回贬

Me®Proc[object][Print)[Name^Freeze]〔Estimate(ForecasHEtats|[Resids]

DependentVariable:PCONS

Method:LeastSquares

Date:06/16/12Time:13:50

Sample:131

Includedobservations:31

VariableCoefficientStd.Errort-StatisticProb.

C282.2434287.26490.9825200.3340

PDINC0.7585110.03692820.540260.0000

R-squared0.935685Meandependentvar5982.476

AdjustedR-squared0.933467S.D.dependentvar1601.762

S.E.ofregression413.1593Akaikeinfocriterion14.94788

Sumsquaredresid4950317.Schwarzcriterion15,04040

Loglikelihood-229.6922Hannan-Quinncriter.14,97804

F-statistic421.9023Durbin-Watsonstat1.481439

Prob(F-statistic)0.000000

3、单击name,输入eqOl,给方程命名

Nametoidentifyobject

24charactersmaximum,16

orfewerrecommended

Displaynameforlabelingtablesandgraphs(optional)

QK][Cancel|

匕联世螯逼丘&川」几£“上工天组的"外3-』兀，一目回国

[viewRProcJobject||Print||Name[Freeze[〔Estimate伍应叔恒矣][Resids)

DependentVariable:PCONS

Method:LeastSquares

Date:06/16/12Time:13:50

Sample:131

Includedobservations:31

VariableCoefficientStd.Errort-StatisticProb.

C282.2434287.26490.9825200.3340

PDINC0.7585110.03692820.540260.0000

R-squared0.935685Meandependentvar5982.476

AdjustedR-squared0.933467S.D.dependentvar1601.762

S.E.ofregression413.1593Akaikeinfocriterion14,94788

Sumsquaredresid4950317.Schwarzcriterion15.04040

Loglikelihood-229.6922Hannan-Quinncriter.14.97804

F-statistic421.9023Durbin-Watsonstat1.481439

Prob(F-statistic)0.000000

(2)1、单击ViewActual,Fitted,ResidualActual,Fitted,ResidualGraph得到因变量的实际值、

拟合值、残差值的折线图。

(3)1、单击Forecast输入“pconsf”得到预测结果。从图中可以看出，平均百分比误差MAPE=5.22,

希尔不等系数TheilIC=0.03,偏差率BP-0,方差率VP=0.02,协变率CP=0.96,从以上预测评价指标可

以看出模型预测精度高。

Forecast:F1

Actual:PCONS

Forecastsample:131

Includedobservations31

RootMeanSquaredError399.6094

MeanAbsoluteError305.3822

MeanAbs.PercentError5.217788

TheilInequalityCoefficient0.032331

BiasProportion0.000000

VarianceProportion0.016618

CovarianceProportion0.983382

(4)

单击ViewCoefficientTestWald-CoefficientRestrictions输入c(2)=0.75得至llWald系数检验结果。

[view][Proc[object][PrinH[NameRFreeze]〔Estimate[Forecastgtats^Resids]

WaldTest:

Equation:EQ01

TestStatisticValuedfProbability

F-statistic0.053123(1.29)0.8193

Chi-square0.05312310.8177

NullHypothesisSummary:

NormalizedRestriction(=0)ValueStd.Err.

-0.75+C(2)0.0085110.036928

Restrictionsarelinearincoefficients.

[View][Proc)[objbcHproperties][Print][Name[Freeze]DefaultLJSI

RESID01

Lastupdated:06/16/12-13:55

Modified:131//eq01.makeresid

1548.3316

2-172.9288

3-279.5766

4458.4026

5-12.16119

6111.4740

7-56.76591

8-447.5075

9131.6327

10-442.4763

11-455.5791

12-121.3674■

13-620.7974

14-537.8198

15-461.5020

16-514.7703

17177.5070

1814.32975

AnmeEARC

Exercise3-2

(1)

1＞打开源文件Exercise文件中的Exercise3-2。

2^单击QuickEquationEstimation,输入investmentcproduct对方进行设定

3、从回归结果中可以看出自变量product解释了因变量均值95%的变化,说明回归方程拟合优度较好,

product的回归系数为3.4说明product每增加一个单位，因变量investment的均值增加3.4个单位，

回归系数的t检验在统计上显著，说明估计回归方程正确。

口直理近E。走赞近遴宜翁f宓i--u区

Mew][Proc)[objecH|PrinH[NameRFreeze]【EstimateRForecasH[stats][Resids]

Dependentvariable:INVESTMENT

Method:LeastSquares

Date:06/16/12Time:14:03

Sample:19811992

Includedobservations:12

VariableCoefficientStd.Errort-StatisticProb.

C-12.262562.482162-4.9402740.0006

PRODUCT3.4098810.25943713.143400.0000

R-squared0.945280Meandependentvar20.02333

AdjustedR-squared0.939808S.D.dependentvar5.032936

S.E.ofregression1.234782Akaikeinfocriterion3.410678

Sumsquaredresid15.24687Schwarzcriterion3.491496

Loglikelihood-18.46407Hannan-Quinncriter.3.380756

F-statistic172.7489Durbin-Watsonstat1.462938

Prob(F-statistic)0.000000

(2)单击ViewActual,Fitted,ResidualActual,Fitted,ResidualGraph得到因变量的实际值，拟

合值，残差值的折线图。

(3)

1单击ViewStabilityTestChowBreakpointTest输入1988,分割点检验结果。从回归结果看

LR检验结果在统计上不显著，所以接受无结构变化的原假设

n昌

Mew|【Proc]|objecH[PrinHMme[Freeze]归stimatv随缸8贪版也RResids]

'ChowBreakpointTest:1988

jNullHypothesis:Nobreaksatspecifiedbreakpoints

1Varyingregressors:Allequationvariables

=EquationSample:19811992

F-statistic1.350461Prob.F(2,8)0.3124

Loglikelihoodratio3.490659Prob.Chi-Square⑵0.1746

WaldStatistic2.700921Prob.Chi-Square(2)0.2591

2、单击ViewStabilityTestChowForecastTest输入1988,得至UChow预测检验结果。回归结

果表明LR检验在0.05的显著性水平下是显著的，所以应拒绝无结构变化的原假设。

□矍崩贽/爸f搭i碗多如二面逐:^^董：・

Mew]|Proc][object][PrinHMmc[Freeze]四imate帆缸由对如区][Resids]

ChowForecastTest:Forecastfrom1988to1992

F-statistic3.546879Prob.F(5,5)0.0955

Loglikelihoodratio18.17329Prob.Chi-Square(5)0.0027

TestEquation:

Dependentvariable:INVESTMENT

Method:LeastSquares

Date:06/16/12Time:14:07

Sample:19811987

Includedobservations:7

VariableCoefficientStd.Errort-StatisticProb.

C-11.532754.391068-2.6264100.0467

PRODUCT3.3096680.5217126.3438610.0014

R-squared0.889489Meandependentvar16.25429

AdjustedR-squared0.867387S.D.dependentvar2.248828

S.E.ofregression0.818934Akaikeinfocriterion2.673329

Sumsquaredresid3.353261Schwarzcriterion2.657875

Loglikelihood-7.356651Hannan-Quinncriter.2.482317

F-statistic40.24457Durbin-Watsonstat2.279734

Prob(F-statistic)0.001437

(4)

l、单击View——StabilityTest——RecursiveEstimates(OLSonly)——CUSUMTest,得到CUSUM检

验结果。

OutputCoefficientdisplaylist

◎覆cursiveResiduals?c(l)c(2)

OCUSUMTest

OCUSUMofSquaresTest

OQne-StepForecastTest

ON-StepForecastTest

ORecursiveCoefficients

□SaveResultsasSeriesOKCancel

2、单击ViewStabilityTestRecursiveEstimates(OLSonly)CUSUMofSquareTest得到

CUSUM的平方检验结果。

第四章非线性模型的回归估计方法

Exercise4-1

(1)

1、打开源文件Exercise文件中的Exercise4-1。

2、点击QuickEquationEstimation,numbercpopulation,得到回归模型。

从回归结果中可以看出，解释变量population解释了因变量number均值76.4%的变化。说明回归方程

拟合优度较好，解释变量的回归系数的t检验在5%的显著性水平下通过检验，说明回归方程正确。

0金川」卫旧岂义匚工"一J」J

Mew]回oc][objecH[PrieH|Name『FreezeJ[EstimateHForeDasHEtats]〔Resids]

DependentVariable:NUMBER

Method:LeastSquares

Date:06/16/12Time:14:15

Sample:120

Includedobservations:20

VariableCoefficientStd.Errort-StatisticProb.

C-481.6892288.5747-1.6692010.1124

POPULATION5.0557630.6621867.6349650.0000

R-squared0.764067Meandependentvar1464.350

AdjustedR-squared0.750959S.D.dependentvar1212.582

S.E.ofregression605.1268Akaikeinfocriterion15,74339

Sumsquaredresid6591211.Schwarzcriterion15.84297

Loglikelihood-155.4339Hannan-Quinncriter.15.76283

F-statistic58.29270Durbin-Watsonstat1.685326

Prob(F-statistic)0.000000

(2)

单击ProcMakeResidualSeries生成残差序列。

□

Residualtype

@Ordinary

OK

Generalized

Nameforresidseries

residO1

卜询]回比][object]|Properties][PrinH[NameJ[Fre8ze]Default[$0代]同血+/・]

RES1

Lastupdated:06/16/12-14:16

Modified:120//eq01.makeresid

11662.684

2-199.8763

3894.9455

4-565.6683

5-350.9619

6-523.2186

7-27.18484

8-18.99905

9-429.2259

10-135.0994

11-409.7648

12-559.1665

13-146.7687

14-250.3897

15590.3358

16-48.14400

17-626.5457

18598.5664

19626.2318

20-81.75002

(3)

分别建立三个新序列seriesw1=1/@abs(resid01)seriesw2=1/@sqrt(population)series

w3=1/population

单击QuickEquationEstimation使用列表形式对方程进行设定。在输入框内输入"numberc

populationOption,选择WeightedLS/TSLS,在weight对话框中分别输入wlw2w3分别得到用

残差序列的绝对值倒数作为权重，用自变量序列平方根作为权重及用自变量序列的倒数作为权重重新

估计的回归方程。

从上述三个回归结果的R平方值及t检验结果判断得出以残差序列绝对值的倒数作为权重进行加权最

小二乘回归得到的结果拟合优度最高，并且t检验显著。

FileEditObjectViewProcQv

seriesw1=1/@sqrt(abs(res1))

Mew][Pro&[objecH[PrinQlName[Freeze]|Estimae随他曰竟版击|[Resids]

DependentVariable:NUMBER

Method:LeastSquares

Date:06/16/12Time:14:36

Sample:120

Includedobservations:20

Weightingseries:W1

VariableCoefficientStd.Errort-StatisticProb.

C-194.3556204.0205-0.9526280.3534

POPULATION4.0898430.5515547.4151240.0000

WeightedStatistics

R-squared0753371Meandependentvar1329.136

AdjustedR-squared0.739669S.D.dependentvar1016.960

S.E.ofregression266.2018Akaikeinfocriterion14.10103

Sumsquaredresid1275541.Schwarzcriterion14.20060

Loglikelihood-139.0103Hannan-Quinncriter.14.12046

F-statistic54.98407Durbin-Watsonstat1.678825

Prob(F-statistic)0.000001

UnweightedStatistics

R-squared0.731070Meandependentvar1464.350

AdjustedR-squared0.716129S.D.dependentvar1212.582

S.E.ofregression646.0579Sumsquaredresid7513035.

Durbin-Watsonstat1.379292

seriesw1=1/@sqrt(abs(res1))

seriesw2=1/@sqrt(population)

SpecificationOptions

LS&TSLSoptions

__neierosKeaasucixy

I__Iconsistentcoefficient

，)White

Range:120-20obsNewey-West

Sample:120-20obs0WeightedLS/TSLS

(notavailablewith

eqO1Weight：w2|

numberARMAoptions

Startingcoefficient

res1OLS/TSLS

resid

[^1BackcastMAterms

Q[噩

[view]|Prod[objecH|PrinH[Name^Freeze][Estimate]〔Forecastgtats^Resids|

Dependentvariable:NUMBER

Method:LeastSquares

Date:06/16/12Time:14:38

Sample:120

Includedobservations:20

Weightingseries:W2

VariableCoefficientStd.Errort-StatisticProb.

C227.7496164.21001.3869410.1824

POPULATION3.2126580.5229306.1435670.0000

WeightedStatistics

R-squared0.677092Meandependentvar1242.009

AdjustedR-squared0.659153S.D.dependentvar544.1256

S.E.ofregression452.4691Akaikeinfocriterion15.16196

Sumsquaredresid3685110.Schwarzcriterion15.26153

Loglikelihood-149.6196Hannan-Quinncriter.15.18139

F-statistic37.74342Durbin-Watsonstat1.584931

Prob(F-statistic)0.000008

UnweightedStatistics

R-squared0.662522Meandependentvar1464.350

AdjustedR-squared0.643773S.D.dependentvar1212.582

S.E.ofregression723.7266Sumsquaredresid9428042.

Durbin-Watsonstat1.260066

W3充当为-晅运izif尹&皿纯拄且更注、沪上先近“五工关以1

IQFileEditObjectViewProcQuickOptionsWindowHelp

View^Proc[object][PrinH〔Name帕eeze][EstimategorecasHEtnts][Resids]

DependentVariable:NUMBER

Method:LeastSquares

rixe£aityojeciviewrrocDate:06/16/12Time:14:42

seriesw1=1/@sqrt(abs(res1))Sample:120

Includedobservations:20

seriesw2=1/@sqrt(population)

Weightingseries:W3

；eriesw3=1/population

Variable