循环在 r 中的多个列中应用 dyplr

Question

I have a dataset

outcome_data_wide_score1 <- outcome_data %>% select(group, community, site, sessions, patientid, score1) %>% arrange(sessions) %>% filter(group == 2 & province == "X") %>% pivot_wider(., names_from =sessions, values_from = c(score1))

I am trying to run these codes get the similar output for different score columns and if possible few lines of codes which will run the codes for different group, community and site by some list or other command

Any help in this regards will be appreciable.

Answer 1

Using dummy data similar to those posted in the question, you can create an index based on group , community and site . Split it into a list, apply the task for all scores and then bind all together. I have added a function to do this:

library(reshape2)
library(tidyverse)
set.seed(123)
#Data
df <- data.frame(group=1,community=c(rep('x',6),rep('y',3)),
                 site=c(rep(c('a'),3),rep(c('b'),3),rep(c('c'),3)),
                 patientid=c(rep(1,3),rep(2,3),rep(3,3)),
                 sessions=rep(c(1,2,3),3),
                 score1=round(runif(9,0,100),0),
                 score2=round(runif(9,0,100),0),
                 score3=round(runif(9,0,100),0),
                 score4=round(runif(9,0,100),0),
                 score5=round(runif(9,0,100),0))
#Create an id by the columns you want
df$id <- paste(df$group,df$community,df$site)
#Now split
List <- split(df,df$id)
#Function to process
myfun <- function(x)
{
  #Filter columns
  y <- x[,names(x)[which(grepl(c('patient|session|score'),names(x)))]]
  #Melt
  z <- melt(y,id.vars = c('patientid','sessions'))
  #Transform
  u <- pivot_wider(z, names_from =c(variable,sessions), values_from = value)
  #Combine and separate
  ids <- x[,'id',drop=F]
  ids <- ids[!duplicated(ids$id),,drop=F]
  idsg <- separate(ids,col = id,sep = ' ',into = c('group','community','site'))
  #Bind
  w <- bind_cols(idsg,u)
  return(w)
}
#Now apply to List
List2 <- lapply(List,myfun)
#Bind all
DF <- do.call(rbind,List2)
rownames(DF)<-NULL

It will produce this, where scores and sessions are separated by _ :

  group community site patientid score1_1 score1_2 score1_3 score2_1 score2_2 score2_3 score3_1 score3_2
1     1         x    a         1       29       79       41       46       96       45       33       95
2     1         x    b         2       88       94        5       68       57       10       69       64
3     1         y    c         3       53       89       55       90       25        4       66       71