CODEWITHSUNDEEP
rtidyr
user3375672
user3375672

Reputation: 3768

R: use tidyr to clean-up data table with structural missing and redundant data

Still trying to get my hands on tidyrpackages. If one has a data set with redundant rows like this: 

require(dplyr)
require(tidyr)
data <-
      data.frame(
        v1 = c("ID1", NA, "ID2", NA),
        v2 = c("x", NA, "xx", NA),
        v3 = c(NA, "z", NA, "zz"),
        v4 = c(22, 22, 6, 6),
        v5 = c(5, 5, 9, 9)) %>%
      tbl_df()

> data
Source: local data frame [4 x 5]

   v1 v2 v3 v4 v5
1 ID1  x NA 22  5
2  NA NA  z 22  5
3 ID2 xx NA  6  9
4  NA NA zz  6  9

Since the id variables v1- v3 is split into redundant rows with many NAs (and therefore the two measurements are also repeated) one would like to get something like this below:

    v1  v2  v3  v4  v5
1   ID1 x   z   22  5
2   ID2 xx  zz  6   9

What would be a general way of getting this using tidyr ? I feel it could be done using gather() but how ?

Upvotes: 0

Views: 329

Answers (2)

akrun
akrun

Reputation: 887048

You may also do

library(dplyr)
data %>% 
     mutate(v3=v3[!is.na(v3)][cumsum(is.na(v3))]) %>%
     na.omit()
#    v1 v2 v3 v4 v5
#1 ID1  x  z 22  5
#2 ID2 xx zz  6  9

Or based on the data showed

 data %>% 
      mutate(v3=lead(as.character(v3))) %>% 
      na.omit()

Upvotes: 2

jazzurro
jazzurro

Reputation: 23574

One way would be like this. Using na.locf() from the zoo package, I replaced NAs in v1. Then, I grouped the data using the variable. I employed na.locf() one more time to take care of v3. Finally, I removed rows with NAs in v2.

library(zoo)
library(dplyr)

mutate(data, v1 = na.locf(v1)) %>%
group_by(v1) %>%
mutate(v3 = na.locf(v3, fromLast = TRUE)) %>%
filter(complete.cases(v2)) %>%
ungroup

#   v1 v2 v3 v4 v5
#1 ID1  x  z 22  5
#2 ID2 xx zz  6  9

Upvotes: 2

Related Questions