Merge rows in data frame using key

Question

If I have a data framework with columns of columns and columns of data, for example

df <- cbind(key=c("Jane", "Jane", "Sam", "Sam", "Mary"), var1=c("a", NA, "a", "a", "c"), var2=c(NA, "b", NA, "b", "d"))

key    var1 var2
"Jane" "a"  NA  
"Jane" NA   "b" 
"Sam"  "a"  NA
"Sam"  "a"  "b" 
"Mary" "c"  "d" 
"Mary" "c"  NA

And I want a framework that concatenates strings by name, rewriting NAs when possible, for example

key    var1 var2
"Jane" "a"  "b"
"Sam"  "a"  "b"
"Mary" "c"  "d"

How can i do this?

Answer 1

library(data.table)
dtt <- as.data.table(df)

dtt[, list(var1=unique(var1[!is.na(var1)])
         , var2=unique(var2[!is.na(var2)]))
    , by=key]

    key var1 var2
1: Jane    a    b
2: Mary    c    d
3:  Sam    a    b

Answer 2

dplyr. , cbind() , , , , , , . . , , , , - .

df <- data.frame(
  key = c("Jane", "Jane", "Sam", "Sam", "Mary"), 
  var1 = c("a", NA, "a", "a", "c"), 
  var2 = c(NA, "b", NA, "b", "d"),
  stringsAsFactors = FALSE
)

library(dplyr)

collapse <- function(x) x[!is.na(x)][1]

df %.% 
  group_by(key) %.%
  summarise(var1 = collapse(var1), var2 = collapse(var2))
# Source: local data frame [3 x 3]
# 
#  key var1 var2
# 1 Mary    c    d
# 2  Sam    a    b
# 3 Jane    a    b