1、精品 料推薦基于 GLM（廣義線性模型）的數(shù)據(jù)分析SAS 里的 GLM應(yīng)用在實(shí)際中比較廣泛， 對(duì)數(shù)據(jù)的分析具有比較強(qiáng)的普適性。趨勢(shì)面回歸分析（ Trend Analysis ） 是以多元回歸分析為理論基礎(chǔ)的一種預(yù)測(cè)與統(tǒng)計(jì)技術(shù)。 它用空間坐標(biāo)法進(jìn)行多項(xiàng)式回歸， 從中估計(jì)出最佳的回歸模型， 因此也被稱為趨勢(shì)面分析， 當(dāng)不知道手中的數(shù)據(jù)呈線性還是非線性相關(guān)時(shí)， 可以采用趨勢(shì)面數(shù)據(jù)分析方法，以便找出擬合數(shù)據(jù)的最佳統(tǒng)計(jì)預(yù)測(cè)模型。本文運(yùn)用 GLM對(duì)一定的數(shù)據(jù)進(jìn)行GLM分析。一、 數(shù)據(jù)與要求此處選取 15 名吧不同程度的煙民的每日飲酒 （啤酒）量與心電圖指標(biāo) （ zb）的對(duì)應(yīng)數(shù)據(jù)。然后設(shè)法建立 zb 與日抽

2、煙量（ X）/ 支和日飲酒量（ y） / 升之間的關(guān)系。序號(hào)組別日抽煙量（ x） / 支日飲酒量（ y）/ 升心電圖指標(biāo)（ zb）113010280212511260313513330414014400514514410622012270721811210822512280922513300102231329011340144101234515420133481642514350184501535519470二、 運(yùn)用 GLM 過(guò)程進(jìn)行趨勢(shì)面分析1. 趨勢(shì)分析的 GLM 程序data beer;input obsn x y zb;cards;01 30 10 28002 25 11 2601精

3、品 料推薦03 35 13 33004 40 14 40005 45 14 41006 20 12 27007 18 11 21008 25 12 28009 25 13 30010 23 13 29011 40 14 41012 45 15 42013 48 16 42514 50 18 45015 55 19 470;proc glm;model zb=x y/p;proc glm;model zb=x y x*x x*y y*y/p;proc glm;model zb=x y x*x*x x*x*y x*y*y y*y*y/p;proc glm;model zb=x y x*x*xx*x

4、*yx*y*yy*y*yx*x*x*xx*x*x*yx*x*y*yx*y*y*yy*y*y*y/p;run;2. 四種分析模型結(jié)果（1）一階趨勢(shì)模型Dependent Variable: zb源變量自由度平方和均值F 值概率值Sum ofSourceDFSquaresMean SquareF ValuePr FModel290615.2099345307.60497127.19 Fx189541.5655889541.56558251.36 F2精品 料推薦x114652.2435114652.2435141.13 |t|Intercept64.0499938033.065399191.940

5、.0766x5.383855650.839475676.41 FModel593330.8358018666.16716107.75 FX189541.5655889541.56558516.86 Fx1965.2913631965.29136315.570.0426y1127.4395437127.43954370.740.4133x*x143.662297243.66229720.250.6277x*y1242.0343234242.03432341.400.2675y*y149.843031649.84303160.290.6047StandardParameterEstimateErr

6、ort ValuePr |t|Intercept-262.7664793109.1074817-2.410.0394x16.06997796.80786202.360.0426y23.539132727.44498670.860.4133x*x0.06387730.12723830.500.6277x*y-1.16510160.9857119-1.180.2675y*y1.16733622.17629820.540.6047-ObservationObservedPredictedResidual1280.0000000279.41687000.58313002260.0000000258.6

7、8145961.31854043330.0000000351.0997183-21.09971834400.0000000388.125128211.87487185410.0000000414.0657505-4.06575056270.0000000255.125602414.87439767210.0000000216.6773768-6.67737688280.0000000279.94178340.05821669300.0000000303.5367795-3.536779510290.0000000295.5572467-5.557246711410.0000000388.125

8、128221.874871812420.0000000419.02805850.971941513425.0000000436.4318573-11.431857314450.0000000453.7554706-3.755470615470.0000000465.43176994.5682301-Sum of Residuals-0.000000Sum of Squared Residuals1559.164195Sum of Squared Residuals - Error SS-0.000000First Order Autocorrelation-0.354205Durbin-Wat

9、son D2.6948084精品 料推薦（3）三階趨勢(shì)模型Dependent Variable: zb源變量自由度平方和均值F 值概率值Sum ofSourceDFSquaresMean SquareF ValuePr FModel693393.4641415565.5773683.21 Fx189541.5655889541.56558478.66 Fx11643.3470811643.3470818.780.0180y1197.474017197.4740171.060.3343x*x*x1105.516422105.5164220.560.4741x*x*y1113.710330113.

10、7103300.610.4580x*y*y1146.610010146.6100100.780.4018y*y*y1173.116161173.1161610.930.3642StandardParameterEstimateErrort ValuePr |t|Intercept-166.007458982.37772231-2.020.0786x11.13825983.757952332.960.0180y15.778434015.357039051.030.3343x*x*x-0.01541320.02052250-0.750.4741x*x*y0.12031870.154323330.7

11、80.4580x*y*y-0.34167860.38595313-0.890.4018y*y*y0.31348940.325876140.960.3642ObservationObservedPredictedResidual1280.0000000281.0906363-1.09063632260.0000000256.04837833.95162173330.0000000351.8935219-21.89352194400.0000000390.57078969.42921045410.0000000409.23096520.76903486270.0000000257.99834901

12、2.00165107210.0000000220.0483966-10.04839665精品 料推薦8280.0000000275.01603684.98396329300.0000000299.47099730.529002710290.0000000295.8228899-5.822889911410.0000000390.570789619.429210412420.0000000420.5758580-0.575858013425.0000000437.4437284-12.443728414450.0000000455.6875798-5.687579815470.000000046

13、3.53108336.4689167-Sum of Residuals-0.000000Sum of Squared Residuals1496.535862Sum of Squared Residuals - Error SS-0.000000First Order Autocorrelation-0.357545Durbin-Watson D2.686333-（4）四階趨勢(shì)模型Dependent Variable: zb源變量自由度平方和均值F 值概率值Sum ofSourceDFSquaresMean SquareF ValuePr FModel1194480.319198589.119

14、9362.900.0029Error3409.68081136.56027Corrected Total1494890.00000R-SquareCoeff VarRoot MSEzb Mean0.9956833.36769511.68590347.0000SourceDFType I SSMean SquareF ValuePr Fx189541.5655889541.56558655.690.0001y11073.644351073.644357.860.0676x*x*x12078.776642078.7766415.220.0299x*x*y1508.85526508.855263.7

15、30.1491x*y*y117.5061417.506140.130.7440y*y*y1173.11616173.116161.270.3421x*x*x*x152.9156652.915660.390.5777x*x*x*y1193.81980193.819801.420.3192x*x*y*y1452.42798452.427983.310.1663x*y*y*y140.3287940.328790.300.6246y*y*y*y1347.36281347.362812.540.2090-SourceDFType III SSMean SquareF ValuePr Fx153.8347

16、35453.83473540.390.5746y118.442245818.44224580.140.73766精品 料推薦x*x*x1707.3985134707.39851345.180.1073x*x*y1688.7276032688.72760325.040.1104x*y*y1669.2155979669.21559794.900.1137y*y*y1614.9897506614.98975064.500.1239x*x*x*x173.525495773.52549570.540.5162x*x*x*y121.572098721.57209870.160.7176x*x*y*y115

17、0.8940383150.89403831.100.3704x*y*y*y1264.7516451264.75164511.940.2581y*y*y*y1347.3628138347.36281382.540.2090StandardParameterEstimateErrort ValuePr |t|Intercept-748.5352475602.9093096-1.240.3026x21.526850134.28557060.630.5746y63.4532525172.66693160.370.7376x*x*x1.11290830.48897822.280.1073x*x*y-7.

18、84664423.4939960-2.250.1104x*y*y17.69195997.99199322.210.1137y*y*y-12.81731806.0398396-2.120.1239x*x*x*x-0.00528950.0072088-0.730.5162x*x*x*y-0.03396280.0854515-0.400.7176x*x*y*y0.42181270.40127851.050.3704x*y*y*y-1.09527330.7866207-1.390.2581y*y*y*y0.84110790.52737831.590.2090ObservationObservedPre

19、dictedResidual1280.0000000280.6428697-0.64286972260.0000000254.91486495.08513513330.0000000336.2353148-6.23531484400.0000000399.845152400000000409.00291000.99709006270.0000000265.56236444.43763567210.0000000212.0079405-2.00794058280.0000000287.4716063-7.47160639300.0000000292.67012457.329875510290.0000000295.8090433-5.809043311410.0000000399.845152410.154847612420.0000000428.1747562-8.174756213425.0000000422.52284782.477152214450.0000000450.5733972-0.573397215470