Recode values based on values in other dataset

Question 1

I want to recode values in one dataset based on values in another dataset. My overall goal is to applyrecode across multiple columns of the dataframe.

Data:

df <- data.frame(  gender=c(1,2,1,2),  condition=c(1,1,2,2))df  gender condition1      1         12      2         13      1         24      2         2

Other dataset:

codes <- data.frame(  gender_values= c("`1`='male', `2`='female'"),  condition_values = c("`1`='exp', `2`='control'"))codes             gender_values         condition_values1 `1`='male', `2`='female' `1`='exp', `2`='control'

Attempt:

df %>%   dplyr::mutate(  gender= dplyr::recode(gender, cat(noquote(codes[1,"gender_values"])), .default = NA_character_))`1`='male', `2`='female'  gender condition1   <NA>         12   <NA>         13   <NA>         24   <NA>         2

Wanted:

  gender condition1   male       exp2 female       exp3   male   control4 female   control

Question 2

In yourcodes, you havec("1='male',2='female'") which is one long, single string. Was this intentional, or is this supposed to be two elements (1="male" and2="female")? Something like,c("1='male'", "2='female'")?

Question 3

You sure that codes has this weird form? Why not log form (i.e. variable_name, variable_vaule, variable_text? Or wide form? Most straight forward solution would be to use a join to bind the "recoded" values to the numbers...

Question 4

Does this answer your question?Recoding values in second data frame based on values in a different data frame

Question 5

@jpsmith I intended a long string so recode can use the values. For some reason, recode is not using the values in the same way as recode(gender,1="male"...).

Question 6

@dario I believe it does not because the value labels for my dataset are in one cell for each variable.

Question 7

If you want to usedplyr::recode, you can exploit the splice operator!!! to help out here and evaluate the values incodes. It helps if you simplify yourcodes data:

codes <- data.frame(  gender_values = c("male", "female"),  condition_values = c("exp", "control"))

Then, for example, on a single column you can do:

dplyr::recode(df$gender, !!!codes$gender_values)# [1] "male"   "female" "male"   "female"

One way to apply it across columns given your example data is to usesapply:

sapply(names(df), function(x) dplyr::recode(df[,x], !!!codes[,paste0(x, "_values")]))#      gender   condition# [1,] "male"   "exp"    # [2,] "female" "exp"    # [3,] "male"   "control"# [4,] "female" "control"

(Note this specific example assumed all column names incodes for variable “x” (indf) is “x_values”, as in your example data)

Also, if you wanted to keep youcodes values exactly as is, you could do:

codes <- data.frame(  gender_values= c("`1`='male', `2`='female'"),  condition_values = c("`1`='exp', `2`='control'"))# single column exampledplyr::recode(df$gender, !!!strsplit(codes$gender_values, ",")[[1]])# [1] "`1`='male'"    " `2`='female'" "`1`='male'"    " `2`='female'"# multiple columnssapply(names(df), function(x) dplyr::recode(df[,x], !!!strsplit(codes[,paste0(x, "_values")], ",")[[1]]))#      gender          condition       # [1,] "`1`='male'"    "`1`='exp'"     # [2,] " `2`='female'" "`1`='exp'"     # [3,] "`1`='male'"    " `2`='control'"# [4,] " `2`='female'" " `2`='control'"

To clean it up a bit, though, you could preprocess the original codes:

codes2 <- lapply(codes[], function(x)  gsub("`|'|=|[[:digit:]]+", "", trimws(unlist(strsplit(x, ",")))))sapply(names(df), function(x) dplyr::recode(df[,x], !!!codes2[[paste0(x, "_values")]]))#      gender   condition# [1,] "male"   "exp"    # [2,] "female" "exp"    # [3,] "male"   "control"# [4,] "female" "control"

I would also advise changing "exp" to "exposure" or "expos" or something else, asexp is a function in R to calculate exponentials, and its good practice not to confuse things!

Question 8

I’d probably need the dput - this is best suited for its own question as it may be valuable to others. Ping me if you create a new one - Good luck!

jpsmith 18.6k6 gold badges25 silver badges48 bronze badges · Accepted Answer · 2023-01-18 16:28:53Z

If you want to usedplyr::recode, you can exploit the splice operator!!! to help out here and evaluate the values incodes. It helps if you simplify yourcodes data:

codes <- data.frame(  gender_values = c("male", "female"),  condition_values = c("exp", "control"))

Then, for example, on a single column you can do:

dplyr::recode(df$gender, !!!codes$gender_values)# [1] "male"   "female" "male"   "female"

One way to apply it across columns given your example data is to usesapply:

sapply(names(df), function(x) dplyr::recode(df[,x], !!!codes[,paste0(x, "_values")]))#      gender   condition# [1,] "male"   "exp"    # [2,] "female" "exp"    # [3,] "male"   "control"# [4,] "female" "control"

(Note this specific example assumed all column names incodes for variable “x” (indf) is “x_values”, as in your example data)

Also, if you wanted to keep youcodes values exactly as is, you could do:

codes <- data.frame(  gender_values= c("`1`='male', `2`='female'"),  condition_values = c("`1`='exp', `2`='control'"))# single column exampledplyr::recode(df$gender, !!!strsplit(codes$gender_values, ",")[[1]])# [1] "`1`='male'"    " `2`='female'" "`1`='male'"    " `2`='female'"# multiple columnssapply(names(df), function(x) dplyr::recode(df[,x], !!!strsplit(codes[,paste0(x, "_values")], ",")[[1]]))#      gender          condition       # [1,] "`1`='male'"    "`1`='exp'"     # [2,] " `2`='female'" "`1`='exp'"     # [3,] "`1`='male'"    " `2`='control'"# [4,] " `2`='female'" " `2`='control'"

To clean it up a bit, though, you could preprocess the original codes:

codes2 <- lapply(codes[], function(x)  gsub("`|'|=|[[:digit:]]+", "", trimws(unlist(strsplit(x, ",")))))sapply(names(df), function(x) dplyr::recode(df[,x], !!!codes2[[paste0(x, "_values")]]))#      gender   condition# [1,] "male"   "exp"    # [2,] "female" "exp"    # [3,] "male"   "control"# [4,] "female" "control"

I would also advise changing "exp" to "exposure" or "expos" or something else, asexp is a function in R to calculate exponentials, and its good practice not to confuse things!

I’d probably need the dput - this is best suited for its own question as it may be valuable to others. Ping me if you create a new one - Good luck!

Movatterモバイル変換

Collectives™ on Stack Overflow

Recode values based on values in other dataset

1 Answer1

1 Comment

Your Answer

Sign up orlog in

Post as a guest

Linked

Related

Hot Network Questions

Subscribe to RSS