Once again i'm struggling with map functions of purrr.
I've got a list of dataframes, all with ID and Name columns.
I want to perform some recoding and then aggregation on rows with some specific values. For that purpose, i've got another dataframe with a vector of ID and newID that I want to replace before doing some aggregration (sum all numeric variable).
I know how to perform this on one df (see II/), but I don't know what test to write in a map_if function to apply thoses operations on all dataframes where column ID includes some values of new newIDdf$ID (here dataframe B and C).
Any ideas ?
## I/ 2 objects
# a list of df
list_df <- list(A = data.frame(ID = c("a", "b", "c", "Z", "Y"),
Name = c("a_name", "b_name", "c_name", "Z_name", "Y_name"),
Var1 = rnorm(5),
Var2 = rnorm(5),
Var3 = rnorm(5)),
B = data.frame(ID = c("a", "b", "z1", "z2", "z3"),
Name = c("a_name", "b_name", "z1_name", "z2_name", "z3_name"),
Var1 = rnorm(5),
Var2 = rnorm(5)),
C = data.frame(ID = c("y1", "y2", "z1", "z2", "z3"),
Name = c("y1_name", "y2_name", "z1_name", "z2_name", "z3_name"),
Var1 = rnorm(5),
Var2 = rnorm(5)))
# a dataframe of correspondance for aggregation operations
newIDdf <- data.frame(ID = c("y1", "y2", "z1", "z2", "z3"),
IDagr = c("Y", "Y", "Z", "Z", "Z"))
## II/ what I want to do (but on 1 df)
# example on 1 df
On1df <- list_df[["B"]] %>%
mutate(ID = reduce2(newIDdf$ID, newIDdf$IDagr,
.init= ID,
str_replace)) %>%
mutate(Name = case_when(ID == "Z" ~ "Z_name",
ID == "Y" ~ "Y_name",
TRUE ~ Name)) %>%
group_by(ID) %>%
mutate_if(is.numeric, ~list(. = sum(.))) %>%
distinct(ID, .keep_all = TRUE)
## III/ What I really want to achieve
# what if I want to do that simultaneously on df B and C
# I mean applying thoses operations on dataframes
# where column ID includes some values of new newIDdf$ID
list_df_output <- list_df %>% map_if( .p = ~ any(ID %in% newIDdf$ID), ### what test to put here ? (because this doesn't work)
~ mutate(.x, ID = reduce2(newIDdf$ID, newIDdf$IDagr,
.init= ID,
str_replace)) %>%
mutate(.,Name = case_when(ID == "Z" ~ "Z_name",
ID == "Y" ~ "Y_name",
TRUE ~ Name)) %>%
group_by(., ID) %>%
mutate_if(., is.numeric, ~list(. = sum(.))) %>%
distinct(., ID, .keep_all = TRUE) )
Do you want this? I also changed your mutate_at function to the more recent version using across and where:
list_df |>
map_if(~any(.x$ID %in%newIDdf$ID) , ~ .x |>
mutate(ID = reduce2(newIDdf$ID, newIDdf$IDagr,
.init= ID,
str_replace)) %>%
mutate(Name = case_when(ID == "Z" ~ "Z_name",
ID == "Y" ~ "Y_name",
TRUE ~ Name)) %>%
group_by(ID) %>%
mutate(across(where(is.numeric), ~ sum(.))) %>%
distinct(ID, .keep_all = TRUE))
Output:
$A
ID Name Var1 Var2 Var3
1 a a_name 0.1015844 0.6306434 0.5058593
2 b b_name -0.1420690 0.5152645 0.2497879
3 c c_name 0.5841423 1.2883330 0.5297098
4 Z Z_name 1.6645565 0.2307524 -1.0418045
5 Y Y_name -0.1293767 -2.4152871 -0.1935843
$B
# A tibble: 3 × 4
# Groups: ID [3]
ID Name Var1 Var2
<chr> <chr> <dbl> <dbl>
1 a a_name -0.512 -0.119
2 b b_name -2.14 -0.834
3 Z Z_name 0.468 2.54
$C
# A tibble: 2 × 4
# Groups: ID [2]
ID Name Var1 Var2
<chr> <chr> <dbl> <dbl>
1 Y Y_name 1.15 0.162
2 Z Z_name 0.790 2.03