循环遍历data.frame以使用不同的DV和权重运行相同的回归

Question

嗨，我无法通过lapply获取lm来循环浏览不同的DV，我已经在data.frame中预先确定了相应的权重列。

这是一个有效的例子：

require(dplyr)


df <- mtcars %>% mutate_at(
  vars(mpg:drat),
  funs(wt=rnorm(n()))
  )
dvs <- names(select(df,mpg:drat))
res <- list()
res <- lapply(dvs, function(x) {
  formula <- as.formula(paste0(x," ~ wt + qsec + as.factor(vs)"))
  lm(formula,df)
})

但这会返回“找不到对象'mpg_wt'”

res <- lapply(dvs, function(x) {
  formula <- as.formula(paste0(x," ~ wt + qsec + as.factor(vs)"))
  weightvars <- paste0(x,"_wt")
  lm(formula,weights=as.formula(weightvars),df)
})

并返回“变量长度不同（找到'（权重）'）”

res <- lapply(dvs, function(x) {
  formula <- as.formula(paste0(x," ~ wt + qsec + as.factor(vs)"))
  weightvars <- paste0(x,"_wt")
  lm(formula,weights=eval(weightvars),df)
})

Answer 1

只需使用解析并传递weightvars作为text的值。

我希望这有帮助：

res <- lapply(dvs, function(x) {
  formula <- as.formula(paste0(x," ~ wt + qsec + as.factor(vs)"))
  weightvars <- paste0(x,"_wt")
  lm(formula,weights=eval(parse(text=weightvars)),df)
})

或者在as.name中使用eval函数，如下所示：

res <- lapply(dvs, function(x) {
  formula <- as.formula(paste0(x," ~ wt + qsec + as.factor(vs)"))
  weightvars <- paste0(x,"_wt")
  lm(formula,weights=eval(as.name(weightvars)),data=df)
})