根据摘要选择最佳模型（R）

Question

我试图使用分数多项式模型，我得到了以下输出。我不知道如何决定哪一个是最好的。我该怎么看？

Call:
lm(formula = y ~ FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE))

Residuals:
    Min      1Q  Median      3Q     Max 
-27.694  -3.313   1.103   3.400  26.140 

Coefficients: (6 not defined because of singularities)
                                                                       Estimate Std. Error t value Pr(>|t|)
(Intercept)                                                           -70097287  188740046  -0.371    0.710
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^-2             170753     403346   0.423    0.672
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^-1           -3933621    9796171  -0.402    0.688
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^-0.5         21945827   56730458   0.387    0.699
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^0.5         168218120  473584466   0.355    0.722
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^1          -104378366  293650571  -0.355    0.722
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^2           -14874379   49275635  -0.302    0.763
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)x^3             2948966   10676603   0.276    0.782
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)               NA         NA      NA       NA
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^-2        35905      83761   0.429    0.668
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^-1           NA         NA      NA       NA
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^-0.5         NA         NA      NA       NA
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^0.5          NA         NA      NA       NA
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^1      49476323  146274523   0.338    0.735
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^2            NA         NA      NA       NA
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)x^3       -960045    3668249  -0.262    0.794
FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)log(x)^2             NA         NA      NA       NA

Residual standard error: 7.999 on 2236 degrees of freedom
  (433 observations deleted due to missingness)
Multiple R-squared:  0.5023,    Adjusted R-squared:    0.5 
F-statistic: 225.6 on 10 and 2236 DF,  p-value: < 2.2e-16

这是我的代码：

polynomial.model <- lm(y ~ FP(x, p = c(-2, -1, -0.5, 0.5, 1, 2, 3), scaling = TRUE)) 
summary(polynomial.model) 

我使用了mboost包。