根据行将单列拆分为多列

Question

I have a dataset in R that is comprised of a single column containing variables that I ideally would like in multiple columns. The structure of the single column dataframe is this.

A1
200
250
Brand x 
A2
400
300
Brand x
A4
100
320
Brand x2

I would like to split this column in such a way that it ends up in a multi-column frame like this ("|" is purely to denote a column separator):

A1 | 200 | 250 | Brand x  
A2 | 400 | 300 | Brand x1
A4 | 100 | 320 | Brand x2

How could I do this? There is at most times a sequence in the horizontal data - for example: 4 variables - A1,200, 250,Brand x. Naive equivalent would be copying and transpose pasting in Excel, but for a predefined sequence of 4 values. Could anyone please help me with this?

Answer 1

Here's how I would do it:

df2 <- as.data.frame(matrix(df1[,1], byrow=TRUE, ncol = 4))

or, equivalently:

df2 <- as.data.frame(t(matrix(df1[,1],nrow = 4)))

In both cases this yields the desired result:

#> df2
#  V1  V2  V3       V4
#1 A1 200 250  Brand x
#2 A2 400 300  Brand x
#3 A4 100 320 Brand x2

data

df1 <-read.table(text="A1
                       200
                       250
                       'Brand x' 
                       A2
                       400
                       300
                      'Brand x'
                       A4
                       100
                       320
                       'Brand x2'", header=FALSE)

Answer 2

This is not an ellegant solution but should work.

Some explanations:

The first two lines should only provide the dataframe which you usually obtain by reading in your data.

If there is a character string in a column R will transform this column in a factor variable. For this reason I transformed it in line 3 back to a character vector.

With matrix you can rearrange this vector in the shape you want and than you can transform it back to a dataframe (setting stringAsFactors=FALSE to prevent that everything is transformed into factors which would be the default).

However, now all variables are character variables. For this reason you need to encode the variables appropriately.

dat<-c("A1",200,250,"Brand x" ,"A2",400,0300, "Brand x", "A4",100,  320,"Brand x2")
dat<-data.frame(dat)
dat<-as.character(dat[,1])
dat<-matrix(dat, ncol = 4, byrow=TRUE)
dat<-data.frame(dat, stringsAsFactors = FALSE)

dat[] <- lapply(dat, type.convert)

> str(dat)
'data.frame':   3 obs. of  4 variables:
 $ X1: Factor w/ 3 levels "A1","A2","A4": 1 2 3
 $ X2: int  200 400 100
 $ X3: int  250 300 320
 $ X4: Factor w/ 2 levels "Brand x","Brand x2": 1 1 2

> dat
     X1  X2  X3       X4
 1 A1 200 250  Brand x
 2 A2 400 300  Brand x
 3 A4 100 320 Brand x2

Answer 3

Just a hint here - if the sequence always repeats (ie is deterministic), you could read a vector and change the dimensions, something like:

data <- c("A1","200","250","Brand x","A2","400","300","Brand x","A4","100","320","Brand x2")
dim(data) <- c(4,3)
data <- t(data) # transpose
class(data)
data.df <- as.data.frame(data)
class (data.df)

This change the dims of the data to a matrix (since internally vector and matrix are stored the same, it's the dimensions that differ).

When executed, it will print

> class(data)
[1] "matrix"
> class (data.df)
[1] "data.frame"

and the data.df is then a data.frame object, so you can do whatever you need to do with the data (eg change column to be numeric/character/etc) before processing the data.

Answer 4

If it's always 4 values the loop below did the work for me:

df <- read.csv("df.csv", sep = ";", header = FALSE)


new.df <- data.frame()
j <- 1
i <- 1
while(i < length(df[,1])-1){

    temp.df <- data.frame()

    temp.df[j,1] <- df[i,1]
    temp.df[j,2] <- df[i + 1, 1]
    temp.df[j,3] <- df[i + 2, 1]
    temp.df[j,3] <- df[i + 3, 1]

    new.df <- rbind(new.df, temp.df)

    j <- j + 1
    i <- i + 4
}
na.omit(new.df)

it's not entirely optimized but it does the job! Hope it works for you.