ranell - 1 month ago 7
R Question

Reshaping datas in a specific form

I've datas as follows, it is a but in reality i've few

experiment
, it is simplified dataset:

DF=structure(list(theoric = c("E", "E", "F", "F", "F"), observed = c("E",
"E", "F", "F", "E"), experiment = c("RO(2)", "RO(2)", "RO(2)", "RO(2)",
"RO(2)")), .Names = c("theoric", "observed", "experiment"), row.names = 2:6, class = "data.frame")

Now my datas has the following form:

theoric observed experiment
2 E E RO(2)
3 E E RO(2)
4 F F RO(2)
5 F F RO(2)
6 F E RO(2)

Adn I want it to be reshaped as follows :

2 3 4 5 6
RO(2) theoric E E F F F
RO(2) observed E E F F E

What is the easiest way to do it ? I really have no idea how to do this. I tried

meltR <- melt(DF, id="experiment")

But i'm lost all correspondance between
theoric
and
observed
. Thanks a lot

EDIT : full dataset:

DF=structure(list(theoric = c("E", "E", "F", "F", "F", "E", "F",
"F", "F", "F", "F", "E", "E", "E", "E"), observed = c("E", "E",
"F", "F", "E", "F", "F", "F", "F", "F", "F", "E", "E", "E", "F"
), experiment = c("RO", "RO", "RO", "RO", "RO", "MO", "MO", "MO",
"MO", "MO", "MO", "EL", "EL", "EL", "EL")), .Names = c("theoric",
"observed", "experiment"), row.names = c(2L, 3L, 4L, 5L, 6L,
24L, 25L, 26L, 27L, 28L, 29L, 21L, 22L, 23L, 13L), class = "data.frame")

output:

col2 col1.2 col1.3 col1.4 col1.5 col1.6 col1.24 col1.25 col1.26
1 RO theoric E E F F F <NA> <NA> <NA>
6 MO theoric <NA> <NA> <NA> <NA> <NA> E F F
12 EL theoric <NA> <NA> <NA> <NA> <NA> <NA> <NA> <NA>
16 RO observed E E F F E <NA> <NA> <NA>
21 MO observed <NA> <NA> <NA> <NA> <NA> F F F
27 EL observed <NA> <NA> <NA> <NA> <NA> <NA> <NA> <NA>
col1.27 col1.28 col1.29 col1.21 col1.22 col1.23 col1.13
1 <NA> <NA> <NA> <NA> <NA> <NA> <NA>
6 F F F <NA> <NA> <NA> <NA>
12 <NA> <NA> <NA> E E E E
16 <NA> <NA> <NA> <NA> <NA> <NA> <NA>
21 F F F <NA> <NA> <NA> <NA>
27 <NA> <NA> <NA> E E E F

EL
ouput

RO theoric E E F F F
RO observed E E F F E
MO theoric E F F F F
MO observed F F F F F
EL theoric E E E E
EL observed E E E F

Based on the expected output, we may need to create a column with row.names. Create a new dataset ('df2'), by unlisting the first two columns, replicating the 'experiment' column, and a rownames column. Then use reshape from base R to convert the 'long' format to 'wide'.

df2 <- data.frame(col1 = unlist(DF[1:2], use.names=FALSE),
col2 = paste( rep(DF\$experiment, 2),
rep(colnames(DF)[1:2], each = nrow(DF))), col3 = rep(row.names(DF), 2))

reshape(df2, idvar = "col2", direction="wide", timevar = "col3")
#             col2 col1.2 col1.3 col1.4 col1.5 col1.6
#1  RO(2) theoric      E      E      F      F      F
#6 RO(2) observed      E      E      F      F      E

Or using melt/dcast from data.table. Convert the 'data.frame' to 'data.table' keeping the row names (setDT(DF, keep.row.names = TRUE)), melt it to 'long' format, paste the 'experiment' and 'variable' column, and then dcast from 'long' to 'wide' format.

library(data.table)
dcast(melt(setDT(DF, keep.rownames = TRUE), id.var = c("rn", "experiment"))[,
experiment := paste(experiment, variable)], experiment~rn, value.var = "value")
#       experiment 2 3 4 5 6
#1: RO(2) observed E E F F E
#2:  RO(2) theoric E E F F F

Update

Using the new dataset,

library(data.table)#v1.9.7+
dcast(melt(setDT(DF), id.var = "experiment"), paste(experiment,
variable)~rowid(experiment, variable), value.var="value", fill="")
#    experiment 1 2 3 4 5 6
#1: EL observed E E E F
#2:  EL theoric E E E E
#3: MO observed F F F F F F
#4:  MO theoric E F F F F F
#5: RO observed E E F F E
#6:  RO theoric E E F F F
Source (Stackoverflow)