library(mgcv)

##1 读取数据
trainset <- read.delim(file="F:/07.txt",header=TRUE)
testset <- read.delim(file="F:/2012/07.txt",header=TRUE)
sst<-trainset[,1]
stg<-trainset[,2]
ssh<-trainset[,3]
sss<-trainset[,4]
CPUE<-trainset[,5]
mgam<-gam(log(CPUE+1)~s(sst)+s(ssh)+s(sss)+s(stg),data=trainset,model=T)
summary(mgam)

##2运行结果
Family: gaussian
Link function: identity

Formula:
log(CPUE + 1) ~ s(sst) + s(ssh) + s(sss) + s(stg)

Parametric coefficients:
            Estimate Std. Error t value Pr(>|t|)   
(Intercept)  6.11369    0.07581   80.65   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Approximate significance of smooth terms:
         edf Ref.df     F  p-value   
s(sst) 7.603  8.505 5.034 5.51e-06 ***
s(ssh) 5.255  6.430 3.018  0.00568 **
s(sss) 2.139  2.692 4.305  0.00740 **
s(stg) 2.032  2.584 0.904  0.44626   
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

R-sq.(adj) =   0.21   Deviance explained = 24.7%
GCV = 2.2236  Scale est. = 2.1147    n = 368

##3 提问
GAM模型的运行结果能否得到log（CPUE+1）与s(sst)、s(ssh)、s(sss)、s(stg)之间的关系，怎么看？
谢谢大家！！