[quote]
引用第2楼谢益辉于2007-01-08 16:04发表的“”:
里面怎么有那么多-1?……
做了个回归,发现效果不好哦:)[/quote]
我也做了,结论是所有变量基本上都找不到"响应"。用交叉验证线性回归,结果如下
<br />
Validation selection criteria:<br />
(Intercept) dat1[, -5]1 dat1[, -5]2 dat1[, -5]3 dat1[, -5]4 cv<br />
[1,] 1 0 0 0 0 0.002421<br />
[2,] 1 0 1 0 0 0.002591<br />
[3,] 1 1 0 0 0 0.002699<br />
[4,] 1 0 0 0 1 0.002805<br />
[5,] 1 0 0 1 0 0.002902<br />
[6,] 1 1 1 0 0 0.002932<br />
[7,] 1 0 1 0 1 0.003019<br />
[8,] 1 0 1 1 0 0.003093<br />
[9,] 1 1 0 0 1 0.003143<br />
[10,] 1 1 0 1 0 0.003194<br />
[11,] 1 0 0 1 1 0.003345<br />
[12,] 1 1 1 0 1 0.003439<br />
[13,] 1 1 1 1 0 0.003458<br />
[14,] 1 0 1 1 1 0.003600<br />
[15,] 1 1 0 1 1 0.003700<br />
[16,] 1 1 1 1 1 0.004059<br />
[17,] 0 0 0 0 1 0.024530<br />
[18,] 0 1 0 0 0 0.024910<br />
[19,] 0 0 0 1 0 0.025540<br />
[20,] 0 0 1 0 0 0.025570<br />
<br />
Printed the first 20 best models <br />
仅对因素2做线性回归,结果如下
<br />
Residuals:<br />
Min 1Q Median 3Q Max <br />
-0.094667 -0.039333 0.001333 0.037833 0.063333 <br />
<br />
Coefficients:<br />
Estimate Std. Error t value Pr(>|t|) <br />
(Intercept) 0.141667 0.009029 15.690 1.89e-14 ***<br />
dat1[, 2] -0.015667 0.013544 -1.157 0.258 <br />
---<br />
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 <br />
<br />
Residual standard error: 0.04692 on 25 degrees of freedom<br />
Multiple R-Squared: 0.0508, Adjusted R-squared: 0.01284 <br />
F-statistic: 1.338 on 1 and 25 DF, p-value: 0.2583 <br />