The code below works fine, but I'd like to create another variable called JOV after SPV. This variable would have some condition as follows:

If I have "Category","Week" and "DTT" in group_cols, do:

  SPV %>% filter(date2 == dmda, Category == CategoryChosse, DTT==DTest)

If I have "Category" and "Week" in group_cols, do:

 SPV %>% filter(date2 == dmda, Category == CategoryChosse)

If I only have "Category" in group_cols, do:

 SPV %>% filter(date2 == dmda)

Executable code below

library(dplyr)
library(tidyverse)
library(lubridate)

df1 <- structure(
  list(date1= c("2021-06-28","2021-06-28","2021-06-28","2021-06-28"),
       date2 = c("2021-06-23","2021-06-24","2021-06-30","2021-07-01"),
       DTT= c("Hol","Hol","Hol",0),
       Week= c("Wednesday","Thursday","Wednesday","Thursday"),
       Category = c("ABC","FDE","ABC","FDE"),
       DR1 = c(4,1,1,2),
       DR01 = c(4,1,2,3), DR02= c(4,2,0,2),DR03= c(9,5,0,1),
       DR04 = c(5,4,3,2),DR05 = c(5,4,0,2)),
  class = "data.frame", row.names = c(NA, -4L))

dmda<-"2021-07-01"
CategoryChosse<-"FDE"
DTest<-"Hol"
Wk<-"Thursday"

Dx<-subset(df1,df1$date2<df1$date1)

x<-Dx %>% select(starts_with("DR0"))

x<-cbind(Dx, setNames(Dx$DR1 - x, paste0(names(x), "_PV")))

PV<-select(x, date2,Week, Category, DTT, DR1, ends_with("PV"))

group_cols <-
  if (any(PV$DTT == DTest & PV$Week == Wk, na.rm = TRUE)) {
    c("Category", "Week", "DTT")
  } else if (any(PV$Week == Wk & PV$Category == CategoryChosse & PV$DTT != DTest, na.rm=TRUE)) {
    c("Category", "Week")
  } else {
    "Week"
  }

med <- PV %>%
  group_by(across(all_of(group_cols))) %>%
  summarize(across(ends_with("PV"), median),.groups = 'drop')

SPV <- df1 %>%
  inner_join(med, by = group_cols) %>%
  mutate(across(matches("^DR0\\d $"), ~.x   
                  get(paste0(cur_column(), '_PV')),
                .names = '{col}_{col}_PV')) %>%
  select(date1:Category, DR01_DR01_PV:last_col())

CodePudding user response：

Try:

SPV %>% 
  filter(
    date2 == dmda,
    !("Category" %in% group_cols) | Category == CategoryChosse,
    !all(c("Category", "DTT") %in% group_cols) | DTT == DTest
  )

That's a literal translation of your conditions. However, if I'm reading it right, it can be simplified a little with

SPV %>% 
  filter(
    date2 == dmda,
    !("Category" %in% group_cols) | Category == CategoryChosse,
    !("DTT" %in% group_cols) | DTT == DTest
  )

if you ever imagine allowing "DTT" and not "Category" in your group_cols. (This works even if that will never happen.)

CodePudding user response：

You can store the different cases in a list and just pull the elements as needed.

1. Change this function to return vector and list indices.

filter_condition <-
  if (any(PV$DTT == DTest & PV$Week == Wk, na.rm = TRUE)) {
    1:3 # if you had options out of order, you could have something like c(1, 3)
  } else if (any(PV$Week == Wk & PV$Category == CategoryChosse & PV$DTT != DTest, na.rm=TRUE)) {
    1:2
  } else {
    1
  }

2. Create the vectors and lists for grouping and filtering.

group_cols <- c("Week", "Category", "DTT")

filter_opts <- rlang::exprs(date2 == dmda,
                            Category == CategoryChosse,
                            DTT == DTest)

3. Change your group_by() to pull from the variable based on which case you are in.

med <- PV %>%
  group_by(across(all_of(group_cols[filter_condition]))) %>%
  summarize(across(ends_with("PV"), median),.groups = 'drop')

4. Same code as before.

  inner_join(med, by = group_cols) %>%
  mutate(across(matches("^DR0\\d $"), ~.x   
                  get(paste0(cur_column(), '_PV')),
                .names = '{col}_{col}_PV')) %>%
  select(date1:Category, DR01_DR01_PV:last_col())

5. Apply the filter conditions based on which case you are in.

SPV %>% 
  filter(!!!filter_opts[filter_condition])

As mentioned in another answer, you can shorten this since it seems like date2 is always desired. But this draws up a bit of a framework for things like this.