Restructuring complicated and large dataframe - dataframe

Question

I have a large dataframe with 41040 obs. and 20 variables.
Here I will simplify the mock data set so it's easier to understand the question.
What I have:
rm(list = ls())
variable <- rep(c('var1', 'var1_2', 'var1_3', 'var1_4'), 5)
group <- as.factor(rep(c('county1', 'county2', 'county3', 'county4'), 5))
year <- rep(c(2000:2004), 4)
month <- c(rep(1:12, 1), 1:8)
value1 <- sample(1:10000, 20)
value2 <- sample(1:10000, 20)
value3 <- sample(1:10000, 20)
mydata <- data.frame(variable, group, year, month, value1, value2, value3)
head(mydata)
variable group year month value1 value2 value3
1 var1 county1 2000 1 4848 4759 6029
2 var1_2 county2 2001 2 7624 3486 6745
3 var1_3 county3 2002 3 4612 9155 4266
4 var1_4 county4 2003 4 1496 2420 9451
5 var1 county1 2004 5 6739 4312 5577
6 var1_2 county2 2000 6 5127 5030 5479
What i want from this, is get another data.frame where values won't be messed up across counties, years or months, but each column will represent one variable from the variable column. To clarify, on the same example I am looking for the quickest way to get this:
var1 <- c(t(mydata[1, 5:7]))
var1_2 <- c(t(mydata[2, 5:7]))
var1_3 <- c(t(mydata[3, 5:7]))
var1_4 <- c(t(mydata[4, 5:7]))
group2 <- rep('county1', 3)
year2 <- rep(2000, 3)
month2 <- rep(1, 3)
mydata2 <- data.frame(group2, year2, month2, var1, var1_2, var1_3, var1_4)
head(mydata2)
group2 year2 month2 var1 var1_2 var1_3 var1_4
county1 2000 1 4848 7624 4612 1496
county1 2000 1 4759 3486 9155 2420
county1 2000 1 6029 6745 4266 9451
After all values for county1, year 2000 and month 1 are written, I want it to go to month 2, year 2000 and county1, than month 3 etc. After all months are done I want year 2001 for county 1 etc, and in the end moving to county2.
I tried various ways with melt(), dcast(), stack(), unstack(), gather() and spread() with no success.

Answer 1

I did it, not super-elegantly though. I just divided the original data.frame into new data.frames with selecting the first 4 variables and than alternating the following variables which needed to be cast. Like this:
res <- select(mydata, c(1:4, 5)) # i changed this 5 to 6, than to 7 etc.
base <- dcast(res, group + year + month ~ variable, value.var = 'value1')
after I did this for each column, I used cbind to create a new, casted dataframe:
cbind(base, var1_2[ , 5:14], var1_3[ , 6:14])
It works, although I would still like to see a nicer way to do this automatically in one or two lines.