Reshape data frames: wide and long format
Using stack() for data frames with a simple structure
Wide format -> long format
set.seed(123)
Nj <- 4
cond1 <- sample(1:10, Nj, replace=TRUE)
cond2 <- sample(1:10, Nj, replace=TRUE)
cond3 <- sample(1:10, Nj, replace=TRUE)
dfTemp <- data.frame(cond1, cond2, cond3)
(res <- stack(dfTemp, select=c("cond1", "cond3"))) values ind
1 3 cond1
2 8 cond1
3 5 cond1
4 9 cond1
5 6 cond3
6 5 cond3
7 10 cond3
8 5 cond3str(res)'data.frame': 8 obs. of 2 variables:
$ values: int 3 8 5 9 6 5 10 5
$ ind : Factor w/ 2 levels "cond1","cond3": 1 1 1 1 2 2 2 2Long format -> wide format
unstack(res) cond1 cond3
1 3 6
2 8 5
3 5 10
4 9 5res$IVnew <- factor(sample(rep(c("A", "B"), Nj), 2*Nj, replace=FALSE))
res$DVnew <- sample(100:200, 2*Nj)
head(res) values ind IVnew DVnew
1 3 cond1 B 189
2 8 cond1 A 169
3 5 cond1 A 163
4 9 cond1 A 197
5 6 cond3 B 198
6 5 cond3 B 168unstack(res, DVnew ~ IVnew) A B
1 169 189
2 163 198
3 197 168
4 151 155Using reshape() for more complex data frames
One within variable
Simulate data
Nj <- 2
P <- 2
Q <- 3
id <- 1:(P*Nj)
DV_t1 <- round(rnorm(P*Nj, -1, 1), 2)
DV_t2 <- round(rnorm(P*Nj, 0, 1), 2)
DV_t3 <- round(rnorm(P*Nj, 1, 1), 2)
IVbtw <- factor(rep(c("A", "B"), Nj))
(dfWide <- data.frame(id, IVbtw, DV_t1, DV_t2, DV_t3)) id IVbtw DV_t1 DV_t2 DV_t3
1 1 A -1.56 0.70 -0.03
2 2 B 0.79 -0.47 0.27
3 3 A -0.50 -1.07 0.37
4 4 B -2.97 -0.22 -0.69idL <- rep(id, Q)
DVl <- c(DV_t1, DV_t2, DV_t3)
IVwth <- factor(rep(1:3, each=P*Nj))
IVbtwL <- rep(IVbtw, times=Q)
dfLong <- data.frame(id=idL, IVbtw=IVbtwL, IVwth=IVwth, DV=DVl)
dfLong[order(dfLong$id), ] id IVbtw IVwth DV
1 1 A 1 -1.56
5 1 A 2 0.70
9 1 A 3 -0.03
2 2 B 1 0.79
6 2 B 2 -0.47
10 2 B 3 0.27
3 3 A 1 -0.50
7 3 A 2 -1.07
11 3 A 3 0.37
4 4 B 1 -2.97
8 4 B 2 -0.22
12 4 B 3 -0.69Wide format -> long format
resLong <- reshape(dfWide, varying=c("DV_t1", "DV_t2", "DV_t3"),
direction="long", idvar=c("id", "IVbtw"),
v.names="DV", timevar="IVwth")
resLong[order(resLong$id), ] id IVbtw IVwth DV
1.A.1 1 A 1 -1.56
1.A.2 1 A 2 0.70
1.A.3 1 A 3 -0.03
2.B.1 2 B 1 0.79
2.B.2 2 B 2 -0.47
2.B.3 2 B 3 0.27
3.A.1 3 A 1 -0.50
3.A.2 3 A 2 -1.07
3.A.3 3 A 3 0.37
4.B.1 4 B 1 -2.97
4.B.2 4 B 2 -0.22
4.B.3 4 B 3 -0.69resLong$IVwth <- factor(resLong$IVwth)
all.equal(dfLong, resLong, check.attributes=FALSE)[1] TRUELong format -> wide format
reshape(dfLong, v.names="DV", timevar="IVwth", idvar=c("id", "IVbtw"),
direction="wide") id IVbtw DV.1 DV.2 DV.3
1 1 A -1.56 0.70 -0.03
2 2 B 0.79 -0.47 0.27
3 3 A -0.50 -1.07 0.37
4 4 B -2.97 -0.22 -0.69Two within variables
Simulate data
Nj <- 4
id <- 1:Nj
t_11 <- round(rnorm(Nj, 8, 2), 2)
t_21 <- round(rnorm(Nj, 13, 2), 2)
t_31 <- round(rnorm(Nj, 13, 2), 2)
t_12 <- round(rnorm(Nj, 10, 2), 2)
t_22 <- round(rnorm(Nj, 15, 2), 2)
t_32 <- round(rnorm(Nj, 15, 2), 2)
dfW <- data.frame(id, t_11, t_21, t_31, t_12, t_22, t_32)Wide format -> long format
(dfL1 <- reshape(dfW, varying=list(c("t_11", "t_21", "t_31"),
c("t_12", "t_22", "t_32")),
direction="long", timevar="IV1", idvar="id",
v.names=c("IV2-1", "IV2-2"))) id IV1 IV2-1 IV2-2
1.1 1 1 9.68 9.39
2.1 2 1 8.31 9.24
3.1 3 1 5.72 8.61
4.1 4 1 10.51 9.58
1.2 1 2 13.85 12.47
2.2 2 2 12.41 19.34
3.2 3 2 14.79 17.42
4.2 4 2 14.76 12.75
1.3 1 3 14.64 14.19
2.3 2 3 14.38 14.07
3.3 3 3 14.11 16.56
4.3 4 3 12.88 14.83dfL2 <- reshape(dfL1, varying=c("IV2-1", "IV2-2"),
direction="long", timevar="IV2",
idvar=c("id", "IV1"), v.names="DV")
head(dfL2) id IV1 IV2 DV
1.1.1 1 1 1 9.68
2.1.1 2 1 1 8.31
3.1.1 3 1 1 5.72
4.1.1 4 1 1 10.51
1.2.1 1 2 1 13.85
2.2.1 2 2 1 12.41Long format -> wide format
dfW1 <- reshape(dfL2, v.names="DV", timevar="IV1",
idvar=c("id", "IV2"), direction="wide")dfW2 <- reshape(dfW1, v.names=c("DV.1", "DV.2", "DV.3"),
timevar="IV2", idvar="id", direction="wide")
all.equal(dfW, dfW2, check.attributes=FALSE)[1] TRUEUseful packages
Package reshape2 provides functions melt() and cast() for an integrated approach to reshaping data frames.
Get the article source from GitHub
R markdown - markdown - R code - all posts
top