CODEWITHSUNDEEP
rtidyverse
Karthik S
Karthik S

Reputation: 11584

Getting an error while trying to input arbitrary number of grouping variables and summary variables using tidy dots in R

I am trying to use tidy dots to build a function that I can use to summarize the data based on the requirement. I need to group by variables using different metrics based on the requirement.

Sample Data:

> dput(d)
structure(list(Branch = c("MDU", "KPA", "OMR", "TVM", "KPA", 
"MDU", "MDU", "MDB", "MDU", "OMR", "MDU", "AMB", "OMR", "OMR", 
"GDY", "TVM", "KLM", "EKM", "MDU", "TVM", "KPA", "TVM", "GDY", 
"EKM", "TVM", "KLM", "TVM", "TVM", "AMB", "TVM"), InvoiceDate = structure(c(18475, 
18475, 18475, 18475, 18475, 18475, 18475, 18475, 18475, 18475, 
18475, 18475, 18475, 18475, 18475, 18475, 18475, 18475, 18475, 
18475, 18475, 18475, 18475, 18475, 18475, 18475, 18475, 18475, 
18475, 18475), class = "Date"), LabourRev = c(1093, 300, 1, 2637.86, 
255, 624, 634.75, 3645, 570, 5250, 900, 1500, 423, 1614, 999, 
1140, 2914, 1640, 0, 2574, 255, 1046, 1999, 2637, 1355, 1144, 
0, 1355, 1999, 3240.5), PartsRev = c(1111.98, 990, 0, 16555.64, 
178, 2857.5, 31.74, 3265.03, 28.5, 1942.18, 45, 75, 21.15, 441.71, 
49.95, 3919.5, 10652.72, 4877.58, 0, 16616.53, 1108.11, 1000.49, 
99.95, 2889.68, 0, 3127.34, 0, 0, 99.95, 3618.73), TotalAmt = c(2204.98, 
1290, 1, 19193.5, 433, 3481.5, 666.49, 6910.03, 598.5, 7192.18, 
945, 1575, 444.15, 2055.71, 1048.95, 5059.5, 13566.72, 6517.58, 
0, 19190.53, 1363.11, 2046.49, 2098.95, 5526.68, 1355, 4271.34, 
0, 1355, 2098.95, 6859.23)), row.names = c(NA, -30L), class = c("tbl_df", 
"tbl", "data.frame"))
>

structure:

> str(d)
Classes ‘tbl_df’, ‘tbl’ and 'data.frame':   30 obs. of  5 variables:
 $ Branch     : chr  "MDU" "KPA" "OMR" "TVM" ...
 $ InvoiceDate: Date, format: "2020-08-01" "2020-08-01" "2020-08-01" "2020-08-01" ...
 $ LabourRev  : num  1093 300 1 2638 255 ...
 $ PartsRev   : num  1112 990 0 16556 178 ...
 $ TotalAmt   : num  2205 1290 1 19194 433 ...

For example, I may group by just Branch or month of Invoice Date or by both.

I tried writing the below function for which I am getting the error as mentioned below:

mult_grouping <- function(df, ..., ...){
   grping_vars <- enquos(..., .named = 1)
   summary_vars <- enquos(..., .named = 1)
    
   smry_sum <- map(summary_vars, function(var){
      expr(sum(!!var, na.rm = 1))
   })
    
   names(smry_sum) <- paste0('sum_',names(smry_sum))
    
   df %>% 
     group_by(!!!grping_vars) %>% 
     summarise(!!!smry_sum, Counts = n())
}

Error: repeated formal argument '...' on line 1

Are we supposed to use tidy dots just once in a function? Could someone let me know how to correct the code.

Upvotes: 1

Views: 47

Answers (1)

Ronak Shah
Ronak Shah

Reputation: 389012

You don't need complicated non-standard evaluation if you can pass column names as string, across handles string column names.

library(dplyr)

mult_grouping <- function(df, group_vars, summary_vars) {
  
  df %>%
    mutate(InvoiceDate = lubridate::month(InvoiceDate)) %>%
    group_by(across(all_of(group_vars))) %>%
    summarise(across(all_of(summary_vars), sum), 
              Counts = n())
}
  1. Single group column, single value column
d %>% mult_grouping("Branch", "LabourRev")

# Branch LabourRev Counts
#  <chr>      <dbl>  <int>
#1 AMB        3499       2
#2 EKM        4277       2
#3 GDY        2998       2
#4 KLM        4058       2
#5 KPA         810       3
#6 MDB        3645       1
#7 MDU        3822.      6
#8 OMR        7288       4
#9 TVM       13348.      8
  1. Multiple group column single value columns.
d %>% mult_grouping(c("Branch", "InvoiceDate"), "LabourRev")

# Branch InvoiceDate LabourRev Counts
#  <chr>        <dbl>     <dbl>  <int>
#1 AMB              8     3499       2
#2 EKM              8     4277       2
#3 GDY              8     2998       2
#4 KLM              8     4058       2
#5 KPA              8      810       3
#6 MDB              8     3645       1
#7 MDU              8     3822.      6
#8 OMR              8     7288       4
#9 TVM              8    13348.      8
  1. Multiple group column, multiple value column
d %>% mult_grouping(c("Branch", "InvoiceDate"), c("LabourRev", "PartsRev"))

#  Branch InvoiceDate LabourRev PartsRev Counts
#  <chr>        <dbl>     <dbl>    <dbl>  <int>
#1 AMB              8     3499      175.      2
#2 EKM              8     4277     7767.      2
#3 GDY              8     2998      150.      2
#4 KLM              8     4058    13780.      2
#5 KPA              8      810     2276.      3
#6 MDB              8     3645     3265.      1
#7 MDU              8     3822.    4075.      6
#8 OMR              8     7288     2405.      4
#9 TVM              8    13348.   41711.      8

To have a flexible date column we can pass it as a separate argument in the function.

mult_grouping <- function(df, group_vars, summary_vars, date_cols = NULL) {
  if(is.null(date_cols)) {
  df %>%
    group_by(across(all_of(group_vars))) %>%
    summarise(across(all_of(summary_vars), sum), 
              Counts = n())
  } else {
    df %>%
      mutate(date_col = lubridate::month(.data[[date_cols]])) %>%
      group_by(across(c(all_of(group_vars), date_col))) %>%
      summarise(across(all_of(summary_vars), sum), 
                Counts = n())
  }
}

and call them as :

d %>% mult_grouping("Branch", "LabourRev")

d %>% mult_grouping("Branch", c("LabourRev", "PartsRev"))

d %>% mult_grouping("Branch", c("LabourRev", "PartsRev"), "InvoiceDate")

Upvotes: 1

Related Questions