Using reshape where there are multiple values at each time point

Question

I'm trying to reshape a longitudinal dataset containing visual measurements for the left and right eyes of several individuals over a one year period. I need to end up with a data.frame() with the headings 'patient','month','re','le' (where 're' means 'right eye' and 'le' means 'left eye')

My data are currently in the format:

 patient','re_month1','le_month1','re_month2','le_month2'....'le_month12'

I know I could use the reshape() function to sort the data if I only had one piece of data per time point. If I were just working with 'patient','month1','month2' etc, I could use the following:

 reshape(dframe,idvar = 'patient',v.names = 'vision',
                       varying = 2:13,direction = "long")

...But how do I do this when there are two pieces of data (or more) at each time point?

you can use the varying argument to separate the left and right, by passing the column indices separately.. `varying=list(c(2,4), c(3,5))`, and so on for all the right and left indices. — user20650, Jan 20 '16 at 10:17

akrun · Accepted Answer · 2016-01-20T10:40:59.237

We can use melt from data.table and specify the measure columns with the patterns argument. The patterns can take multiple regex/fixed column names.

 library(data.table)
 melt(setDT(dframe), id.var="patient", 
             measure = patterns("^re_", "^le_"))
 #    patient variable value1 value2
 #1:       1        1     20     21
 #2:       2        1     25     18
 #3:       3        1     23     22
 #4:       1        2     18     29
 #5:       2        2     22     19
 #6:       3        2     25     24

data

 dframe <- data.frame(patient=1:3, re_month1 = c(20, 25,
  23), le_month1= c(21, 18, 22), re_month2=c(18, 22, 25),
  le_month2= c(29, 19, 24))

Using reshape where there are multiple values at each time point

1 Answers1

data