用列折叠行数据（字符、数字、因子等）

Question

I am trying to collapse this data but I am having trouble. The dataset is huge - more than 100 columns, and over 1,000 rows.

This is an example of the dataset of how it looks like:

https://i.stack.imgur.com/8iZq7.png

I need to be able to collapse the rows together. I cannot add the values inside Lab together because it'll be greater than 1.

I have tried multiple codes, and it doesn't work because it doesn't take into consideration that I have character, numeric, and timestamps in my dataframe.

These are the codes that I have tried with the errors:

COLLAPSE6 <- setDT(TRIALBJH4)[, lapply(.SD, function(x)
                      {x <- unique(x[!is.na(x)])
                       if(length(x) == 1) as.character(x)
                       else if(length(x) == 0) NA_character_
                       else collapse=","}),
             by=ID]

This just added a comma into the columns (considered as multiple) when I need it to either say 0, 1, or NA

COLLAPSE3 %>%
  group_by(ID) %>%
  summarise_all(funs(list(na.omit)))

This just replaced the other columns not listed in the group_by with funs(list(na.omit) - it even replaced the values with it

bjh_sti_merge1 <- bjh_sti_merg6 %>% group_by (ID) %>%
  summarise_each(funs(max(., na.rm = TRUE)))

This doesn't work - it freezes R for me, and I always have to force quit it

bjh_sti_merg10 <- bjh_sti_merg6 %>% group_by (ID) %>%
  summarise(AGE = max(AGE, na.rm=TRUE),
            LAB1 = max(LAB1, na.rm=TRUE),
            LAB3 = max(LAB3, na.rm=TRUE))

This one doesn't work - it just takes the first row of the duplicated ones (I can't use this because sometimes the first row is NA, and the third row could have 1 in the column) - Also, this seems to freeze R when I have more than 20 columns in it

xx <-function(x) x[!is.na(x)]

bjh_sti_merg7 %>% 
  group_by(EPIC_MRN) %>%
  summarise_all(funs(xx))

This doesn't work: it says: Error: Problem with 'summarise()' input 'LAB1'. x Input 'LAB1' must be size 0 or 1, not 2.

I want the end result to have 1 row per ID. The code needs to work for all columns (character, numeric, timestamps, factors, etc.). and something that doesn't freeze RStudio for me. I was always recommended summarise_each, but that kept freezing my laptop (I tried to let it run, it ran for over 2 hours and nothing) and yes, I have uploaded tidyverse, data.table, and dplyr

This also needs to accept NA as well!

I would like the dataset to look like: https://i.stack.imgur.com/yBehQ.png

Answer 1

See if this doesn't work, might take some time to run:

plyr::ddply(df, plyr::.(ID), function(x){
  res <- x[1,]
  if(ncol(x) == 1) return(res)
  for (i in 1:ncol(x)) {
    if(class(x[,i]) != "numeric") next()
    res[,i] <- max(x[,i], na.rm=T)
  }
  return(res)
})

Answer 2

This task should be straightforward. It is not clear to me though how you wish to summarize the AGE, TIME, LAB1 and LAB2 columns. For simplicity sake I have used max(col, na.rm = TRUE) .

library(dplyr)
library(tibble)

data <- tibble(
  ID = c(1, 1, 1, 2, 2, 3, 4, 5, 5, 6, 6, 7),
  SEX = c("M", "M", "M", "F", "F", "M", "M", "F", "F", "M", "M", "F"),
  AGE = c(30, 30, 30, 22, 22, 55, 90, 87, 87, 23, 23, 45),
  TIME = as.POSIXct(rep("02/19/2019 12:00", 12), format = "%m/%d/%Y %H:%M", tz = ""),
  LAB1 = c(0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1, 0),
  LAB2 = c(1, 0, 1, 1, 0, 1, 1, 0, 0, 1, 0, 1)
)

data <- data %>%
  group_by(ID, SEX) %>%
  summarize(AGE = max(AGE, na.rm = TRUE),
            TIME = max(TIME, na.rm = TRUE),
            LAB1 = max(LAB1, na.rm = TRUE),
            LAB2 = max(LAB2, na.rm = TRUE))

With this result:

> data
# A tibble: 7 x 6
# Groups:   ID [7]
     ID SEX     AGE TIME                 LAB1  LAB2
  <dbl> <chr> <dbl> <dttm>              <dbl> <dbl>
1     1 M        30 2019-02-19 12:00:00     1     1
2     2 F        22 2019-02-19 12:00:00     1     1
3     3 M        55 2019-02-19 12:00:00     1     1
4     4 M        90 2019-02-19 12:00:00     1     1
5     5 F        87 2019-02-19 12:00:00     0     0
6     6 M        23 2019-02-19 12:00:00     1     1
7     7 F        45 2019-02-19 12:00:00     0     1