如何执行按 2 个因素分组的多重回归并创建包含 N 和 R 平方的文件？

Question

I am running into some problems again and hope that someone can help me. I am doing research on the effect of ELI on ROS for firms and if the pandemic has an effect on this. For this research, my supervisor for my thesis has asked me to do a regression analysis per year grouped by industries (NAICS) and I am at a loss as to how to do this. I have firms in 46 different industries (NAICS) and 11 years of firm data per firm (2010-2020). Now I would like to run a regression ROS ~ ELI + ELI*Pandemic , for all industries for each year and then capture the resulting N (number of firms per industry) and R-squared in one file. The image below is an example of what I am trying to achieve:

I hope that someone can help me because I am at an absolute loss and I can't seem to find a similar question/answer on SO.

Here is the dput(head()) as an example. NAICS is the industry.

df <- structure(list(NAICS = c(315, 315, 315, 315, 315, 315), 
        Year = c(2010, 2011, 2012, 2013, 2014, 2015), 
        Firm = c("A", "A", "A", "A", "A", "A"), 
        ROS = c(0.17, 0.19, 0.29, 0.3, 0.29, 0.25), 
        ELI = c(0.856264428748774, 0.723379402777553, 0.958341156943977, 0.680567730897854, 0.790480861209701, 0.827279134948296), 
        Pandemic = c(0, 0, 0, 0, 0, 0)), 
        row.names = c(NA, -6L), 
        class = c("tbl_df", "tbl", "data.frame"))

Answer 1

Update02

I have made the necessary modifications on my solution after I received the original data set and I don't there will be any other problems.

library(dplyr)
library(tidyr)
library(broom)
library(purrr)


df %>% 
  group_by(NAICS, Year) %>% 
  add_count(name = "N") %>%
  nest(data = !c(NAICS, Year, N)) %>% 
  mutate(models = map(data, ~ lm(ROS ~ ELI + ELI * Pandemic, data = .)),
         glance = map(models, ~ glance(.x)),
         tidied = map(models, ~ tidy(.x))) %>%
  unnest(glance) %>%
  select(NAICS:N, r.squared, tidied) %>%
  unnest(tidied)


# A tibble: 2,024 x 9
# Groups:   NAICS, Year [506]
   NAICS  Year     N r.squared term         estimate std.error statistic     p.value
   <dbl> <dbl> <int>     <dbl> <chr>           <dbl>     <dbl>     <dbl>       <dbl>
 1   315  2010    12     0.122 (Intercept)    0.0959    0.0123     7.83   0.0000143 
 2   315  2010    12     0.122 ELI            0.0189    0.0160     1.18   0.266     
 3   315  2010    12     0.122 Pandemic      NA        NA         NA     NA         
 4   315  2010    12     0.122 ELI:Pandemic  NA        NA         NA     NA         
 5   315  2011    12     0.129 (Intercept)    0.0999    0.0115     8.70   0.00000559
 6   315  2011    12     0.129 ELI            0.0161    0.0132     1.22   0.251     
 7   315  2011    12     0.129 Pandemic      NA        NA         NA     NA         
 8   315  2011    12     0.129 ELI:Pandemic  NA        NA         NA     NA         
 9   315  2012    13     0.594 (Intercept)   -0.486     0.606     -0.802  0.439     
10   315  2012    13     0.594 ELI            2.11      0.526      4.01   0.00205   
# ... with 2,014 more rows