I have a data frame df in R of the form
a,1
a,4
a,2
b,6
b,8
b,4
c,4
c,5
c,2
I want to represent df in the form
a,1,4,2
b,6,8,4
c,4,5,2
What is a faster way to do this transformation in R, especially if my data frame was of a larger size??
Asked
Active
Viewed 243 times
0
-
This is really a [long to wide dupe](https://stackoverflow.com/questions/5890584/how-to-reshape-data-from-long-to-wide-format), but without column names or a sufficiently unique index. It's all fixable, e.g. `library(tidyverse); df %>% set_names(paste0('V', 1:2)) %>% group_by(V1) %>% mutate(var = paste0('X', row_number())) %>% spread(var, V2)` – alistaire Aug 08 '17 at 21:38
1 Answers
1
By using dplyr and reshape2
library(dplyr)
library(reshape2)
dat=dat%>%group_by(V1)%>%dplyr::mutate(id=row_number())
as.data.frame(acast(dat, V1~id,value.var="V2"))
1 2 3
a 1 4 2
b 6 8 4
c 4 5 2
Data input :
dat
V1 V2
1 a 1
2 a 4
3 a 2
4 b 6
5 b 8
6 b 4
7 c 4
8 c 5
9 c 2
EDIT : Timing
library(microbenchmark)
microbenchmark(
acastmethod=acast(dat, a~id,value.var="b"),
dcastmethod=dcast(dat, a ~ id , value.var = "b"),
tidyrmethod=spread(dat, key = id, value = b),
xtabmethod=xtabs(b ~ a + id, data = dat)
)
Unit: milliseconds
expr min lq mean median uq max neval cld
acastmethod 1.872223 2.035528 2.237846 2.210701 2.349068 3.783507 100 a
dcastmethod 3.124578 3.405817 3.626199 3.579038 3.815807 4.887430 100 b
tidyrmethod 4.025684 4.477290 4.765803 4.725326 5.035862 6.140385 100 c
xtabmethod 5.054490 5.529382 6.378615 5.714020 6.047391 61.242200 100 d
BENY
- 317,841
- 20
- 164
- 234
-
What is the variable before and after ~ in the formula in acast referring to? – hearse Aug 09 '17 at 22:09
-
-
ok. Why not use dcast instead of acast? That way need not use as.data.frame. Is acast faster or so? – hearse Aug 09 '17 at 22:12
-
Sure, also why not use reshape itself with id var, time var and values instead of the cast fn.s. I ask in particular as my dataset is very large ;) – hearse Aug 09 '17 at 22:22
-
1Hi, you can check it , also if you do need a faster speed , please try to look at `Rcpp` – BENY Aug 09 '17 at 22:42
-
-
-
1