从原始数据集中删除由Fixest (feols)删除的观测/行

Question

我有一个大型(数百万)观测数据集，并且我使用了feols来运行一个线性模型。这个模型已经从对缺失值的考虑中删除了许多观察结果。我已经恢复了使用$obs_selection删除的行号，但我无法确定如何使用$obs_selection生成的列表从原始数据集中删除删除的观察。

最后，我想删除掉的观测，然后将$residuals加入到原始数据中。

我最初尝试过这样做(通常在下面的代码中指定)：

df[-object$obs_selection]

但是这会生成一个错误“-rows_to_delete中的错误:对一元运算符的无效参数”，并且类似于这个问题的答案：How do you retrieve the estimation sample in R?中的解决方案(以及我得到的错误)。

在下面的样本数据中，由于缺少值，模型中省略了五个观测值。如何使用fake_lm$obs_selection从原始数据集中删除掉的观测？

谢谢!

数据：

structure(list(ExamType = c("A", "B", "C", "D", "E", "F", "G", 
"A", "B", "C", "D", "E", "F", "G", "A", "B", "C", "D", "E", "F", 
"G", "A", "B", "C", "D", "E", "F", "G", "A", "B"), ExamScore = c(1L, 
2L, 2L, 3L, 1L, 4L, 4L, 5L, 2L, 1L, 4L, 3L, 2L, 5L, 1L, NA, 3L, 
2L, 1L, 2L, 5L, 4L, 4L, 3L, 1L, 2L, 5L, 4L, 3L, 1L), State = c("CA", 
"CA", "AL", "AK", "AK", "CA", "AL", "CO", "AL", "CA", "CA", "CA", 
"CO", "CO", "AR", "AR", "AK", "CA", "CA", "CT", "AL", "CA", "AK", 
"CA", "CA", "AL", "AR", "AR", "CA", "CT"), Male = c(1L, 1L, 0L, 
0L, 1L, 0L, 0L, 0L, 1L, 1L, NA, 1L, 1L, 1L, 0L, 0L, 1L, 0L, 1L, 
0L, 0L, 1L, 0L, 0L, 0L, 1L, 1L, 0L, 1L, 1L), White = c(1L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 
0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 1L, 0L, 0L, 0L), Black = c(0L, 
1L, 0L, NA, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 0L, 
0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 0L, 0L, 1L, 0L, 0L), Latinx = c(0L, 
0L, 0L, 0L, 1L, 0L, NA, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 0L, 0L, 
0L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 1L, 0L), X2.Race = c(0L, 
0L, 0L, 1L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 
0L, 0L, 0L, 0L, 0L, 0L, NA, 0L, 0L, 0L, 0L, 0L, 0L)), row.names = c(NA, 
30L), class = "data.frame")

代码：

library(fixest)

fake_lm <- feols(ExamScore ~ Male + White + Black + Latinx + X2.Race | State, fake_data)
summary(fake_lm)

#These are the dropped observations
rows_to_delete <- fake_lm$obs_selection

# I would like to clean them from my dataset (fake_data), but this
# generates the error
fake_data[-rows_to_delete]

 
# Ultimately, once the original dataset only contains those used in the model, I'll add
# residuals as a column in my dataset
fake_data$resid <- fake_lm$residuals

Answer 1

在经历了一些痛苦之后，我想出了答案。

整数的向量列表可以转换为数据，从那时起，这就变成了一个有趣的问题。

从上面重写一些代码..。

library(tidyverse)

fake_data <- fake_data %>% rowid_to_column()

rows_to_delete <- as.data.frame(fake_lm$selection)
row_to_delete$obsRemoved <- rows_to_delete$obsRemoved * -1

colnames(rows_to_delete) <- c("rowid")

clean_fake_data <- anti_join(fake_data,rows_to_delete,by="rowid")

在这里，您可以按需要添加一列残差。