Question

我正在运行模拟，想知道是否有办法使用broom，dplyr，modelr或purrr访问模型内部的“x”估计值。

这给了我我想要的内容，但我不想在最后一段代码中使用[[1]]。

library(tidyverse)
library(purrr)
library(broom)
mod <- function(df) {
  lm(y ~ x, data = df)
}
sim <- tibble(
model = "model1",
mu = 5,             #this is unknown in practice                         
beta = 2.7,         #this is unknown in practice
sigma = 0.15,       #this is unknown in practice
mu_e = 0,
sigma_e = 1
)
sim_dat <- sim %>% 
crossing(replication = 1:10000) %>%
mutate(e = rnorm(mu_e, mu_e),
       x = sample(c(0,1),size=n(),replace = TRUE,prob=c(0.5, 0.5)),
       y = mu+x*beta+e) %>% 
  group_by(model) %>% 
  nest() %>% 
  mutate(model_fit = map(data, mod)) 

broom::tidy(sim_dat$model_fit[[1]]) %>% 
  filter(term=="x") %>% 
  select(estimate)

Answer 1

您可以使用purrr::map_df()：

map_df(sim_dat$model_fit, broom::tidy) %>% 
    filter(term=="x") %>% 
    select(estimate)

您也可以将其放在mutate(model_fit = ...)中，如下所示：

sim_dat <- sim %>% 
    crossing(replication = 1:10000) %>%
    mutate(e = rnorm(mu_e, mu_e),
           x = sample(c(0,1),size=n(),replace = TRUE,prob=c(0.5, 0.5)),
           y = mu+x*beta+e) %>% 
    group_by(model) %>% 
    nest() %>% 
    mutate(model_fit = map(data, mod),

           # you can pipe inside of mutate()

           x_coef = map_dbl(model_fit, ~broom::tidy(.) %>%
               filter(term =="x") %>% 
               select(estimate) %>%
               unlist() ) )

根据您要为x_coef返回的对象类别，您可以使用map_suffix()并且可能删除unlist()我认为dbl有意义

访问R中模型对象内的估计

1 个答案: