[交流] 【求助】这组数据应该如何分析 已有3人参与

请各位帮忙看一下这组数据，s-1~7为说明变量，我想研究目的变量 与说明变量之间的关系，也就是说，哪几个说明变量与目的变量的相关性最高，最能影响目的变量 应该用什么模型分析为好？之前用R的glm模型的高斯分布分析过，也得出了最适合模型，但是从图上来看感觉比较奇怪，点比较分散，没出现线型的分布。不知道这是说明是说明变数与目的变数之间的相关性很低？还是我用的模型不对？请高手帮忙指点一下，非常感谢！！！ 上传的两个图片一个是一部分数据，另一个是AIC~-6.11得最优模型生成的分布图 以下是用r的分析 > step(model) Start: AIC=-1.47 shd ~ kbi + khr + zfk + kwhb + hkkr + iwa + tks Df Deviance AIC - kwhb 1 2.2608 -3.4549 - tks 1 2.2653 -3.3463 - iwa 1 2.2909 -2.7277 2.2601 -1.4726 - hkkr 1 2.3916 -0.3625 - zfk 1 2.4714 1.4440 - kbi 1 2.4897 1.8491 - khr 1 2.7075 6.4621 Step: AIC=-3.45 shd ~ kbi + khr + zfk + hkkr + iwa + tks Df Deviance AIC - tks 1 2.2687 -5.2643 - iwa 1 2.2944 -4.6432 2.2608 -3.4549 - hkkr 1 2.3920 -2.3537 - kbi 1 2.4897 -0.1508 - zfk 1 2.6246 2.7506 - khr 1 2.7076 4.4630 Step: AIC=-5.26 shd ~ kbi + khr + zfk + hkkr + iwa Df Deviance AIC - iwa 1 2.3169 -6.1077 2.2687 -5.2643 - hkkr 1 2.3995 -4.1806 - kbi 1 2.5146 -1.6032 - zfk 1 2.6335 0.9369 - khr 1 2.7118 2.5480 Step: AIC=-6.11 shd ~ kbi + khr + zfk + hkkr Df Deviance AIC 2.3169 -6.1077 - hkkr 1 2.4217 -5.6752 - zfk 1 2.6858 0.0184 - kbi 1 2.8072 2.4500 - khr 1 2.9502 5.1836 Call: glm(formula = shd ~ kbi + khr + zfk + hkkr, family = gaussian, data = dt) Coefficients: (Intercept) kbi khr zfk hkkr 0.2931 7.3872 -0.3329 0.2187 -5.5888 Degrees of Freedom: 54 Total (i.e. Null); 50 Residual Null Deviance: 4.024 Residual Deviance: 2.317 AIC: -6.108 > model<-glm(shd~kbi+khr+zfk+hkkr,data=dt,family=gaussian) > summary(model) Call: glm(formula = shd ~ kbi + khr + zfk + hkkr, family = gaussian, data = dt) Deviance Residuals: Min 1Q Median 3Q Max -0.43642 -0.14177 0.01095 0.11418 0.46965 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.29308 0.08327 3.520 0.000931 *** kbi 7.38722 2.27101 3.253 0.002050 ** khr -0.33293 0.09005 -3.697 0.000542 *** zfk 0.21866 0.07750 2.822 0.006839 ** hkkr -5.58877 3.71679 -1.504 0.138961 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 (Dispersion parameter for gaussian family taken to be 0.04633779) Null deviance: 4.0235 on 54 degrees of freedom Residual deviance: 2.3169 on 50 degrees of freedom AIC: -6.1077 Number of Fisher Scoring iterations: 2 [ Last edited by javeey on 2010-5-8 at 14:25 ]

回复此楼