How to reshape data from long to wide format

Question

I'm having trouble rearranging the following data frame:

set.seed(45)
dat1 <- data.frame(
    name = rep(c("firstName", "secondName"), each=4),
    numbers = rep(1:4, 2),
    value = rnorm(8)
    )

dat1
       name  numbers      value
1  firstName       1  0.3407997
2  firstName       2 -0.7033403
3  firstName       3 -0.3795377
4  firstName       4 -0.7460474
5 secondName       1 -0.8981073
6 secondName       2 -0.3347941
7 secondName       3 -0.5013782
8 secondName       4 -0.1745357

I want to reshape it so that each unique "name" variable is a rowname, with the "values" as observations along that row and the "numbers" as colnames. Sort of like this:

     name          1          2          3         4
1  firstName  0.3407997 -0.7033403 -0.3795377 -0.7460474
5 secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357

I've looked at melt and cast and a few other things, but none seem to do the job.

Answer 1

Using reshape function:

reshape(dat1, idvar = "name", timevar = "numbers", direction = "wide")

Answer 2

The new (in 2014) tidyr package also does this simply, with gather()/spread() being the terms for melt/cast.

Edit: Now, in 2019, tidyr v 1.0 has launched and set spread and gather on a deprecation path, preferring instead pivot_wider and pivot_longer, which you can find described in this answer. Read on if you want a brief glimpse into the brief life of spread/gather.

library(tidyr)
spread(dat1, key = numbers, value = value)

From github,

tidyr is a reframing of reshape2 designed to accompany the tidy data framework, and to work hand-in-hand with magrittr and dplyr to build a solid pipeline for data analysis.

Just as reshape2 did less than reshape, tidyr does less than reshape2. It's designed specifically for tidying data, not the general reshaping that reshape2 does, or the general aggregation that reshape did. In particular, built-in methods only work for data frames, and tidyr provides no margins or aggregation.

Answer 3

You can do this with the reshape() function, or with the melt() / cast() functions in the reshape package. For the second option, example code is

library(reshape)
cast(dat1, name ~ numbers)

Or using reshape2

library(reshape2)
dcast(dat1, name ~ numbers)

Answer 4

Another option if performance is a concern is to use data.table's extension of reshape2's melt & dcast functions

(Reference: Efficient reshaping using data.tables)

library(data.table)

setDT(dat1)
dcast(dat1, name ~ numbers, value.var = "value")

#          name          1          2         3         4
# 1:  firstName  0.1836433 -0.8356286 1.5952808 0.3295078
# 2: secondName -0.8204684  0.4874291 0.7383247 0.5757814

And, as of data.table v1.9.6 we can cast on multiple columns

## add an extra column
dat1[, value2 := value * 2]

## cast multiple value columns
dcast(dat1, name ~ numbers, value.var = c("value", "value2"))

#          name    value_1    value_2   value_3   value_4   value2_1   value2_2 value2_3  value2_4
# 1:  firstName  0.1836433 -0.8356286 1.5952808 0.3295078  0.3672866 -1.6712572 3.190562 0.6590155
# 2: secondName -0.8204684  0.4874291 0.7383247 0.5757814 -1.6409368  0.9748581 1.476649 1.1515627

Answer 5

With tidyr, there is pivot_wider() and pivot_longer() which are generalized to do reshaping from long -> wide or wide -> long, respectively. Using the OP's data:

single column long -> wide

library(tidyr)

dat1 %>% 
    pivot_wider(names_from = numbers, values_from = value)

# # A tibble: 2 x 5
#   name          `1`    `2`    `3`    `4`
#   <fct>       <dbl>  <dbl>  <dbl>  <dbl>
# 1 firstName   0.341 -0.703 -0.380 -0.746
# 2 secondName -0.898 -0.335 -0.501 -0.175

multiple columns long -> wide

pivot_wider() is also capable of more complex pivot operations. For example, you can pivot multiple columns simultaneously:

# create another column for showing the functionality
dat2 <- dat1 %>% 
    dplyr::rename(valA = value) %>%
    dplyr::mutate(valB = valA * 2) 

dat2 %>% 
    pivot_wider(names_from = numbers, values_from = c(valA, valB))

# # A tibble: 2 × 9
#   name       valA_1 valA_2 valA_3 valA_4 valB_1 valB_2 valB_3 valB_4
#   <chr>       <dbl>  <dbl>  <dbl>  <dbl>  <dbl>  <dbl>  <dbl>  <dbl>
#  1 firstName   0.341 -0.703 -0.380 -0.746  0.682 -1.41  -0.759 -1.49 
#  2 secondName -0.898 -0.335 -0.501 -0.175 -1.80  -0.670 -1.00  -0.349

There is much more functionality to be found in the docs.

Answer 6

Using your example dataframe, we could:

xtabs(value ~ name + numbers, data = dat1)

Answer 7

Other two options:

Base package:

df <- unstack(dat1, form = value ~ numbers)
rownames(df) <- unique(dat1$name)
df

sqldf package:

library(sqldf)
sqldf('SELECT name,
      MAX(CASE WHEN numbers = 1 THEN value ELSE NULL END) x1, 
      MAX(CASE WHEN numbers = 2 THEN value ELSE NULL END) x2,
      MAX(CASE WHEN numbers = 3 THEN value ELSE NULL END) x3,
      MAX(CASE WHEN numbers = 4 THEN value ELSE NULL END) x4
      FROM dat1
      GROUP BY name')

Answer 8

Using base R aggregate function:

aggregate(value ~ name, dat1, I)

# name           value.1  value.2  value.3  value.4
#1 firstName      0.4145  -0.4747   0.0659   -0.5024
#2 secondName    -0.8259   0.1669  -0.8962    0.1681

Answer 9

The base reshape function works perfectly fine:

df <- data.frame(
  year   = c(rep(2000, 12), rep(2001, 12)),
  month  = rep(1:12, 2),
  values = rnorm(24)
)
df_wide <- reshape(df, idvar="year", timevar="month", v.names="values", direction="wide", sep="_")
df_wide

Where

• idvar is the column of classes that separates rows
• timevar is the column of classes to cast wide
• v.names is the column containing numeric values
• direction specifies wide or long format
• the optional sep argument is the separator used in between timevar class names and v.names in the output data.frame.

If no idvar exists, create one before using the reshape() function:

df$id   <- c(rep("year1", 12), rep("year2", 12))
df_wide <- reshape(df, idvar="id", timevar="month", v.names="values", direction="wide", sep="_")
df_wide

Just remember that idvar is required! The timevar and v.names part is easy. The output of this function is more predictable than some of the others, as everything is explicitly defined.

Answer 10

There's very powerful new package from genius data scientists at Win-Vector (folks that made vtreat, seplyr and replyr) called cdata. It implements "coordinated data" principles described in this document and also in this blog post. The idea is that regardless how you organize your data, it should be possible to identify individual data points using a system of "data coordinates". Here's a excerpt from the recent blog post by John Mount:

The whole system is based on two primitives or operators cdata::moveValuesToRowsD() and cdata::moveValuesToColumnsD(). These operators have pivot, un-pivot, one-hot encode, transpose, moving multiple rows and columns, and many other transforms as simple special cases.

It is easy to write many different operations in terms of the cdata primitives. These operators can work-in memory or at big data scale (with databases and Apache Spark; for big data use the cdata::moveValuesToRowsN() and cdata::moveValuesToColumnsN() variants). The transforms are controlled by a control table that itself is a diagram of (or picture of) the transform.

We will first build the control table (see blog post for details) and then perform the move of data from rows to columns.

library(cdata)
# first build the control table
pivotControlTable <- buildPivotControlTableD(table = dat1, # reference to dataset
                        columnToTakeKeysFrom = 'numbers', # this will become column headers
                        columnToTakeValuesFrom = 'value', # this contains data
                        sep="_")                          # optional for making column names

# perform the move of data to columns
dat_wide <- moveValuesToColumnsD(tallTable =  dat1, # reference to dataset
                    keyColumns = c('name'),         # this(these) column(s) should stay untouched 
                    controlTable = pivotControlTable# control table above
                    ) 
dat_wide

#>         name  numbers_1  numbers_2  numbers_3  numbers_4
#> 1  firstName  0.3407997 -0.7033403 -0.3795377 -0.7460474
#> 2 secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357

Answer 11

much easier way!

devtools::install_github("yikeshu0611/onetree") #install onetree package

library(onetree)
widedata=reshape_toWide(data = dat1,id = "name",j = "numbers",value.var.prefix = "value")
widedata

        name     value1     value2     value3     value4
   firstName  0.3407997 -0.7033403 -0.3795377 -0.7460474
  secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357

if you want to go back from wide to long, only change Wide to Long, and no changes in objects.

reshape_toLong(data = widedata,id = "name",j = "numbers",value.var.prefix = "value")

        name numbers      value
   firstName       1  0.3407997
  secondName       1 -0.8981073
   firstName       2 -0.7033403
  secondName       2 -0.3347941
   firstName       3 -0.3795377
  secondName       3 -0.5013782
   firstName       4 -0.7460474
  secondName       4 -0.1745357

Answer 12

This works even if you have missing pairs and it doesn't require sorting (as.matrix(dat1)[,1:2] can be replaced with cbind(dat1[,1],dat1[,2])):

> set.seed(45);dat1=data.frame(name=rep(c("firstName","secondName"),each=4),numbers=rep(1:4,2),value=rnorm(8))
> u1=unique(dat1[,1]);u2=unique(dat1[,2])
> m=matrix(nrow=length(u1),ncol=length(u2),dimnames=list(u1,u2))
> m[as.matrix(dat1)[,1:2]]=dat1[,3]
> m
                    1          2          3          4
firstName   0.3407997 -0.7033403 -0.3795377 -0.7460474
secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357

This doesn't work if you have missing pairs and it requires sorting, but it's a bit shorter in case the pairs are already sorted:

> u1=unique(dat1[,1]);u2=unique(dat1[,2])
> dat1=dat1[order(dat1[,1],dat1[,2]),] # not actually needed in this case
> matrix(dat1[,3],length(u1),,T,list(u1,u2))
                    1          2          3          4
firstName   0.3407997 -0.7033403 -0.3795377 -0.7460474
secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357

Here's a function version of the first approach (add as.data.frame to make it work with tibbles):

l2w=function(x,row=1,col=2,val=3,sort=F){
  u1=unique(x[,row])
  u2=unique(x[,col])
  if(sort){u1=sort(u1);u2=sort(u2)}
  out=matrix(nrow=length(u1),ncol=length(u2),dimnames=list(u1,u2))
  out[cbind(x[,row],x[,col])]=x[,val]
  out
}

Or if you only have the values of the lower triangle, you can do this:

> euro=as.matrix(eurodist)[1:3,1:3]
> lower=data.frame(V1=rownames(euro)[row(euro)[lower.tri(euro)]],V2=colnames(euro)[col(euro)[lower.tri(euro)]],V3=euro[lower.tri(euro)])
> lower
         V1        V2   V3
1 Barcelona    Athens 3313
2  Brussels    Athens 2963
3  Brussels Barcelona 1318
> n=unique(c(lower[,1],lower[,2]))
> full=rbind(lower,setNames(lower[,c(2,1,3)],names(lower)),data.frame(V1=n,V2=n,V3=0))
> full
         V1        V2   V3
1 Barcelona    Athens 3313
2  Brussels    Athens 2963
3  Brussels Barcelona 1318
4    Athens Barcelona 3313
5    Athens  Brussels 2963
6 Barcelona  Brussels 1318
7    Athens    Athens    0
8 Barcelona Barcelona    0
9  Brussels  Brussels    0
> l2w(full,sort=T)
          Athens Barcelona Brussels
Athens         0      3313     2963
Barcelona   3313         0     1318
Brussels    2963      1318        0

Or here's another approach:

> rc=as.matrix(lower[-3])
> n=sort(unique(c(rc)))
> m=matrix(0,length(n),length(n),,list(n,n))
> m[rc]=lower[,3]
> m[rc[,2:1]]=lower[,3]
> m
          Athens Barcelona Brussels
Athens         0      3313     2963
Barcelona   3313         0     1318
Brussels    2963      1318        0

Another simple method in base R is to use xtabs. The result of xtabs is basically just a matrix with a fancy class name, but you can make it look like a regular matrix with class(x)=NULL;attr(x,"call")=NULL;dimnames(x)=unname(dimnames(x)):

> x=xtabs(value~name+numbers,dat1);x
            numbers
name                  1          2          3          4
  firstName   0.3407997 -0.7033403 -0.3795377 -0.7460474
  secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357
> str(x)
 'xtabs' num [1:2, 1:4] 0.341 -0.898 -0.703 -0.335 -0.38 ...
 - attr(*, "dimnames")=List of 2
  ..$ name   : chr [1:2] "firstName" "secondName"
  ..$ numbers: chr [1:4] "1" "2" "3" "4"
 - attr(*, "call")= language xtabs(formula = value ~ name + numbers, data = dat1)
> class(x)
[1] "xtabs" "table"
> class(as.matrix(x)) # `as.matrix` has no effect because `x` is already a matrix
[1] "xtabs" "table"
> class(x)=NULL;class(x)
[1] "matrix" "array"
> attr(x,"call")=NULL;dimnames(x)=unname(dimnames(x))
> x # now it looks like a regular matrix
                    1          2          3          4
firstName   0.3407997 -0.7033403 -0.3795377 -0.7460474
secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357
> str(x)
 num [1:2, 1:4] 0.341 -0.898 -0.703 -0.335 -0.38 ...
 - attr(*, "dimnames")=List of 2
  ..$ : chr [1:2] "firstName" "secondName"
  ..$ : chr [1:4] "1" "2" "3" "4"

Normally as.data.frame(x) converts the result of xtabs back to long format, but you can avoid it with class(x)=NULL:

> x=xtabs(value~name+numbers,dat1);as.data.frame(x)
        name numbers       Freq
1  firstName       1  0.3407997
2 secondName       1 -0.8981073
3  firstName       2 -0.7033403
4 secondName       2 -0.3347941
5  firstName       3 -0.3795377
6 secondName       3 -0.5013782
7  firstName       4 -0.7460474
8 secondName       4 -0.1745357
> class(x)=NULL;as.data.frame(x)
                    1          2          3          4
firstName   0.3407997 -0.7033403 -0.3795377 -0.7460474
secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357

This converts data in wide fromat to long format (unlist converts a dataframe to a vector and c converts a matrix to a vector):

w2l=function(x)data.frame(V1=rownames(x)[row(x)],V2=colnames(x)[col(x)],V3=unname(c(unlist(x))))

Answer 13

Came here via a linked question Reshape three column data frame to matrix ("long" to "wide" format). That question is closed, so I writing an alternative solution here.

I found a alternative solution, perhaps useful for someone looking for converting three columns to a matrix. I am referring to decoupleR (2.3.2) package. Below is copied from their site

---

Generates a kind of table where the rows come from id_cols, the columns from names_from and the values from values_from.

Usage

pivot_wider_profile(
data,
id_cols,
names_from,
values_from,
values_fill = NA,
to_matrix = FALSE,
to_sparse = FALSE,
...
)

Answer 14

Using only dplyr and map.

library(dplyr)
library(purrr)
set.seed(45)
dat1 <- data.frame(
  name = rep(c("firstName", "secondName"), each=4),
  numbers = rep(1:4, 2), value = rnorm(8)
)
longer_to_wider <- function(data, name_from, value_from){
  group <- colnames(data)[!(colnames(data) %in% c(name_from,value_from))]
  data %>% group_by(.data[[group]]) %>%
    summarise( name = list(.data[[name_from]]), 
               value = list(.data[[value_from]])) %>%
    {
      d <- data.frame(
        name = .[[name_from]] %>% unlist() %>% unique()
      )
      e <- map_dfc(.[[group]],function(x){
          y <- data_frame(
            x = data %>% filter(.data[[group]] == x) %>% pull(value_from)
          )
          colnames(y) <- x
          y
      })
      cbind(d,e)
    }
}
longer_to_wider(dat1, "name", "value")
#    name          1          2          3          4
# 1  firstName  0.3407997 -0.7033403 -0.3795377 -0.7460474
# 2 secondName -0.8981073 -0.3347941 -0.5013782 -0.1745357