提取 GAM 模型對象

Question

假設我通過這樣做來創建我的 GAM 模型：

a <- runif(10)
b <- runif(10)
gm <- gam(a ~ ns(b, df=2))
plot(gm, all.terms=T, shade=T)

然后我得到以下情節：

我需要的是訪問gm的元素，以便獲取包含以紅色突出顯示的點的列表或數據框。

通過查看summary(gm)和names(gm)的輸出，我找不到包含此類數據的字段。

> summary(gm)
Family: gaussian 
Link function: identity 

Formula:
a ~ ns(b, df = 2)

Parametric coefficients:
               Estimate Std. Error t value Pr(>|t|)   
(Intercept)      0.5390     0.1524   3.536  0.00952 **
ns(b, df = 2)1   0.4935     0.4242   1.163  0.28284   
ns(b, df = 2)2  -0.2203     0.2603  -0.846  0.42529   
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1


R-sq.(adj) =  -0.0585   Deviance explained = 17.7%
GCV = 0.077126  Scale est. = 0.053988  n = 10

和

> names(gm)
 [1] "coefficients"      "residuals"         "fitted.values"     "family"            "linear.predictors"
 [6] "deviance"          "null.deviance"     "iter"              "weights"           "prior.weights"    
[11] "df.null"           "y"                 "converged"         "sig2"              "edf"              
[16] "edf1"              "hat"               "R"                 "boundary"          "sp"               
[21] "nsdf"              "Ve"                "Vp"                "rV"                "mgcv.conv"        
[26] "gcv.ubre"          "aic"               "rank"              "gcv.ubre.dev"      "scale.estimated"  
[31] "method"            "smooth"            "formula"           "var.summary"       "cmX"              
[36] "model"             "control"           "terms"             "pred.formula"      "pterms"           
[41] "assign"            "xlevels"           "offset"            "df.residual"       "min.edf"          
[46] "optimizer"         "call" 

Answer 1

檢查utils::str （而不是使用summary ）-它為您提供對象的結構。

我認為gm$model就是你要找的。

gm$model
            a ns(b, df = 2).1 ns(b, df = 2).2
1  0.69342149      0.07841860     -0.05184526
2  0.23538533      0.52006793      0.20238728
3  0.47125666      0.24808303     -0.15840080
4  0.04405890      0.00000000      0.00000000
5  0.54696387      0.34211652      0.77302788