#这是个简易的lasso， 里面有几个参数，像family，应该自主 ？？函数 去调一下

#加载包
library(haven)
#导入数据
R <- read_dta("C:/Users/XXX/Desktop/R_cat.dta")
data <- R
#X变量都是连续变量
x <- as.matrix(data[,1:29])
y <- data[,30]   #第30列是Y变量
x <- data.matrix(x)
y <- data.matrix(y)
alpha1_fit <- glmnet(x,y,alpha=1,family="binomial")  #??glmnet看下family可以选择的
plot(alpha1_fit,xvar="lambda",label=TRUE)

alpha1.fit <- cv.glmnet(x,y,type.measure = "auc",alpha=1,family="binomial")
plot(alpha1.fit)
print(alpha1.fit)

coef(alpha1_fit,s=alpha1.fit$lambda.1se)

#如果X为分类变量

library(haven)
R <- read_dta("C:/Users/xxx/Desktop/lasso2.dta")
data <- R

#如果X为分类变量
xfactors <- model.matrix(group ~ agegroup+bmifz+occupation+
                                  brush_rate+denture+hypertention+diabetes
                                 ,data=R)[,-1]
x <- as.matrix(data.frame(xfactors))

y <- data[,22]
y <- data.matrix(y)

library(glmnet)
alpha1_fit <- glmnet(x,y,alpha=1,family="binomial")
plot(alpha1_fit,xvar="lambda",label=TRUE)
set.seed(12345)
alpha1.fit <- cv.glmnet(x,y,type.measure = "auc",alpha=1,family="binomial")
plot(alpha1.fit)
print(alpha1.fit)

coef(alpha1_fit,s=alpha1.fit$lambda.1se)