Error in reshaping and concatenating data frame

Question

I have a data frame:

  out.new_cost out1.new_cost out2.new_cost out3.new_cost out4.new_cost out5.new_cost
1     11049.18      11056.08      11948.41      11048.89      11049.18      11056.14

I would like each of these values to be a separate row. I would then like to add the following information as a column for each row:

alphas <- c(.005, .0005, .00005, .005, .0005, .00005) 

thresholds <- c(.0000001, .0000001, .0000001, .0000001, .0000001, .0000001)

iterations <- c(200000, 200000, 200000, 2000000, 2000000, 2000000)

I tried to start by looking here and using:

reshape(costs, idvar = new_cost, direction = "long")

AND

reshape(costs, direction = "long")

but this returns the error:

Error in reshape(costs, direction = "long") : no 'reshapeWide' attribute, must specify 'varying'

What am I doing wrong and how can I fix it?

Dave2e · Accepted Answer · 2017-09-13T20:22:10.480

2

The gather function in the tidyr package will do the trick.

df<-read.table(header=TRUE, text="out.new_cost out1.new_cost out2.new_cost out3.new_cost out4.new_cost out5.new_cost
    11049.18      11056.08      11948.41      11048.89      11049.18      11056.14")

alphas <- c(.005, .0005, .00005, .005, .0005, .00005) 
thresholds <- c(.0000001, .0000001, .0000001, .0000001, .0000001, .0000001)
iterations <- c(200000, 200000, 200000, 2000000, 2000000, 2000000)

library(tidyr)
df<-gather(df)
#rename first column
names(df)[1]<-"cost"
#remove everything after the .
df$cost<-gsub("\\..*" , "", df$cost )

#add the extra columns
answer<-cbind(df, alphas, thresholds, iterations)

For this type of problem the tidyr package is a better tool than base R, but if the case is simply to change a single row into a column format a simple solution is to transpose such as t(df), then continue with renaming and cbind commands.

edited Sep 13 '17 at 20:22

answered Sep 13 '17 at 20:03

Dave2e

22,192
18
42
50

is there a way to do this without using a package? – zsad512 Sep 13 '17 at 20:05
and how can I change the identifier to "cost" with values "out, out1, out2, out3, out4" etc. – zsad512 Sep 13 '17 at 20:08
For the second question, rename the column name, the parse the column to remove everything before the underscore. – D.sen Sep 13 '17 at 20:17

score 1 · Answer 2 · answered Sep 13 '17 at 21:27

I hope this can help you (R base only)

#Create data frame
costs = data.frame(out.new_cost=11049.18,
out1.new_cost=11056.08,
out2.new_cost=11948.41,
out3.new_cost=11048.89,
out4.new_cost=11049.18,
out5.new_cost=11056.14)

#Create variable with colnames
costs.n = colnames(costs)
#Reshape costs to costs.rshp and saving the colnames to times column
costs.rshp = reshape(costs, direction="long", varying=list(costs.n),     v.names="new_cost",times=costs.n)
#Set the values of new columns
alphas <- c(.005, .0005, .00005, .005, .0005, .00005) 
thresholds <- c(.0000001, .0000001, .0000001, .0000001, .0000001, .0000001)
iterations <- c(200000, 200000, 200000, 2000000, 2000000, 2000000)
#Assign new columns 
costs.rshp$alphas = alphas
costs.rshp$thresholds = thresholds
costs.rshp$iterations = iterations

Error in reshaping and concatenating data frame

2 Answers2