R-Facet_Wrap every n columns together possible ? ggplot2

Question

I am writing a Shiny App based on Cointegration.

After I find the cointegration vectors I plot all the mean reverting series using ggplot2. The way I do it is I put all these cointegration series in a dataframe and I melt it then I facet wrap based on the variable.

Until here I had no problem.

Now I want to add the moving average and the moving standard deviation to the plot of each cointegration series but I can't find a way to have each series with it's MA and STD alone.

I know the problem with what I have right now is with the way I am facet_wraping but I am not an experienced user so I do not know how to solve it. Now I have:

I want the Cointegration Series 1, MA series1 and Std Series 1 to be on the same graph and have different colours and the same for Cointegration Series 2.... The code I use is:

 m= melt(DF, id.vars = "Date") # melt the df

ggplot(m, aes(x = Date, y = value, colour = variable, group=variable)) +geom_line() + facet_wrap( ~ variable) # plot

Is there a way to wrap the first 3 columns together ? Should I melt in a different way ?? Do I need to explain more to be more clear ?

EDIT: Basically in Lehman's terms suppose I have a dataframe df composed of the columns:

Date, S1, MA1, STD1, S2, MA2, STD2

I want my x axis to be Date, and I want to have S1, MA1, STD1 on a plot and S2, MA2 and STD2 on another plot using ggplot2. As id I were facet wraping S1,MA1, STD1 together and then S2, MA2, STD2 also together

Suppose this is the DataFrame:

        Date           S1           M1          ST1         S2          M2        ST2
1 02/12/1999 -0.000217052  0.002862195 -0.002390842  2.2025825 -1.17258213 -0.3057015
2 03/12/1999 -0.004882038 -0.015920939 -0.007014382 -0.4040079 -0.69496488  0.1000167
3 06/12/1999 -0.001445954  0.005077610  0.000000000  1.1573779 -1.64268166 -0.3847015
4 07/12/1999 -0.000907952 -0.005403168  0.002027728 -0.3165827 -1.16466940 -1.0113501
5 08/12/1999  0.000881220  0.011375226 -0.013524439  0.5154455 -0.51725208  1.0942177
6 09/12/1999  0.001043752  0.013400502  0.017033342  1.9368608 -0.05587143  0.5203337

Is it clear enough now ?

Answer 1

I want my x axis to be Date, and I want to have S1, MA1, STD1 on a plot and S2, MA2 and STD2 on another plot using ggplot2.

If you reshape your data as follows:

         Date  S variable        value
1  02/12/1999 S1    value -0.000217052
2  03/12/1999 S1    value -0.004882038
3  06/12/1999 S1    value -0.001445954
4  07/12/1999 S1    value -0.000907952
5  08/12/1999 S1    value  0.000881220
6  09/12/1999 S1    value  0.001043752
7  02/12/1999 S2    value  2.202582500
8  03/12/1999 S2    value -0.404007900
9  06/12/1999 S2    value  1.157377900
10 07/12/1999 S2    value -0.316582700
11 08/12/1999 S2    value  0.515445500
12 09/12/1999 S2    value  1.936860800
13 02/12/1999 S1     mavg  0.002862195
14 03/12/1999 S1     mavg -0.015920939
15 06/12/1999 S1     mavg  0.005077610
16 07/12/1999 S1     mavg -0.005403168
17 08/12/1999 S1     mavg  0.011375226
18 09/12/1999 S1     mavg  0.013400502
19 02/12/1999 S2     mavg -1.172582130
20 03/12/1999 S2     mavg -0.694964880
21 06/12/1999 S2     mavg -1.642681660
22 07/12/1999 S2     mavg -1.164669400
23 08/12/1999 S2     mavg -0.517252080
24 09/12/1999 S2     mavg -0.055871430
25 02/12/1999 S1   stddev -0.002390842
26 03/12/1999 S1   stddev -0.007014382
27 06/12/1999 S1   stddev  0.000000000
28 07/12/1999 S1   stddev  0.002027728
29 08/12/1999 S1   stddev -0.013524439
30 09/12/1999 S1   stddev  0.017033342
31 02/12/1999 S2   stddev -0.305701500
32 03/12/1999 S2   stddev  0.100016700
33 06/12/1999 S2   stddev -0.384701500
34 07/12/1999 S2   stddev -1.011350100
35 08/12/1999 S2   stddev  1.094217700
36 09/12/1999 S2   stddev  0.520333700

Then you can get the plot you want as follows:

ggplot(df, aes(x=Date)) + geom_line(aes(y=value, group=variable, colour=variable)) + facet_wrap(~ S)

Which results in:

Reshaping the data:

There are various packages available for reshaping data including tidyr, reshape2, etc. I'm still learning myself how to use these but this is the convoluted way I reshaped it. I'm sure there are folks that could do this more efficiently. :)

# read the original data
df <- read.table("/tmp/table.dat")

# separate s1 and s2
df.s1 <- df[,1:4]
df.s2 <- df[,c(1,5:7)]

# convert s1 and s2 to long form
df.s1.melted <- melt(df.s1, id.vars=c("Date","M1","ST1"))
df.s2.melted <- melt(df.s2, id.vars=c("Date","M2","ST2"))

# rename columns of s1 and s2 so they can be combined with rbind
s1 <- setNames(df.s1.melted, c("Date", "mavg", "stddev", "S", "value"))
s2 <- setNames(df.s2.melted, c("Date", "mavg", "stddev", "S", "value"))

# combine s1 and s2 with rbind then convert mavg, stddev and value to long form
df <- melt(rbind(s1, s2), id.vars=c("Date", "S"))

# plot it
ggplot(df, aes(x=Date)) + geom_line(aes(y=value, group=variable, colour=variable)) + facet_wrap(~ S)