Plot several graphs using ggplot() and facet_grid()

Question

I am wondering in how to plot several graphs in one screen using ggplot() and facet_grid() because I really need to repeat this process several times for different statistical variables.

I have two data frame contenting observations and another one with predictions. They are both matrix of 550 x 76.

Data frame 1:

Observations    x1   x2   x3  x5  x6  x7  x8  x9  x10  x11  x13 .... x75
Observation1   -0.1 0.05 0.1 0.2 0.3 0.04 0.3 -0.1 0.02 0.02 -0.2 ....0.8
Observation2   -0.3 0.05 0.1 0.1 0.3 0.03 0.3 -0.1 0.03 0.02 -0.2 ....0.6
Observation3   -0.2 0.05 0.1 0.4 0.3 0.02 0.3 -0.1 0.01 0.01 -0.2 ....0.1
Observation550 -0.1 0.05 0.8 0.4 0.3 0.02 0.7 -0.1 0.01 0.01 -0.2 ....0.1

Data frame 2:

Predictions    x1   x2   x3  x5  x6  x7  x8  x9  x10  x11  x13 .... x75
Prediction1   -0.1 0.05 0.1 0.2 0.3 0.04 0.3 -0.1 0.02 0.02 -0.2 ....0.8
Prediction2   -0.3 0.05 0.1 0.1 0.3 0.03 0.3 -0.1 0.03 0.02 -0.2 ....0.6
Prediction3   -0.2 0.05 0.1 0.4 0.3 0.02 0.3 -0.1 0.01 0.01 -0.2 ....0.1
Prediction550 -0.1 0.05 0.8 0.4 0.3 0.02 0.7 -0.1 0.01 0.01 -0.2 ....0.1

It seems that I have to create only one data frame or list with these two matrix inside in order to use ggplot().

I have done this but using the standard plot of r.

Thanks in advance

Answer 1

I think this is what you want.

library(ggplot2)
library(reshape2)

# Fake some data
set.seed(1234)
nc <- 15
nr <- 20  
onms <- sprintf("Observation%d",1:nr)
pnms <- sprintf("Prediction%d",1:nr)
cnames <- sprintf("x%d",1:nc)

odf <- data.frame(Observations=onms)
pdf <- data.frame(Predictions=pnms)

for(i in 1:nc){
  vk1 <- 0.01*rnorm(nr)
  odf[[cnames[i]]] <- round(cumsum(vk1),3)
  vk2 <- 0.02*rnorm(nr)
  pdf[[cnames[i]]] <- round(cumsum(vk1) + cumsum(vk2),3)
}

# This is the data we need
head(odf)
head(pdf)

# Now change the pred. colnames so they don't collide with the obs. colnames
newpnames <- sprintf("p_x%d",1:nc)
names(pdf) <- c("series",newpnames)
names(odf)[1] <- "series"

# Merge the data into a long format
modf <- melt(odf,id.vars="series",measure.vars=cnames)
mpdf <- melt(pdf,id.vars="series",measure.vars=newpnames)   
mdf <- rbind(modf,mpdf)

# Now extract the fields we need into new columns
mdf$x <- as.numeric(gsub(".*[A-Za-z]","",mdf$variable))
mdf$frame <- as.numeric(gsub(".*[A-Za-z]","",mdf$series))
frameNames <- sprintf("Frame:%d",1:max(mdf$frame))
mdf$frame <- factor(sprintf("Frame:%d",mdf$frame),levels=frameNames)
mdf$kind <- substr(mdf$series,1,3)

# Finally plot it
ggplot(mdf) + geom_line(aes(x=x,y=value,color=kind)) + facet_wrap( ~ frame )

# ecdf version
ggplot(mdf,aes(x=value,color=kind)) + stat_ecdf(geom="step") + facet_wrap( ~ frame )

Note that the head statements after I fake the data give you this, so this is close to what you are starting from I believe:

> head(odf)
  Observations     x1     x2     x3     x4     x5    x6     x7     x8     x9
1 Observation1 -0.012  0.014 -0.002 -0.002 -0.008 0.005  0.001 -0.010  0.001
2 Observation2 -0.009  0.004 -0.003 -0.010 -0.011 0.012  0.005 -0.005  0.002
3 Observation3  0.002 -0.005 -0.017  0.011 -0.015 0.014 -0.006 -0.012 -0.003
4 Observation4 -0.022 -0.008 -0.019  0.018 -0.017 0.021  0.001 -0.004 -0.019
5 Observation5 -0.018 -0.017 -0.010  0.037 -0.013 0.024  0.008 -0.012 -0.019
6 Observation6 -0.013 -0.027 -0.003  0.037 -0.007 0.031  0.011 -0.009 -0.026
    x10    x11    x12    x13   x14    x15
1 0.009 -0.012  0.005 -0.007 0.015 -0.007
2 0.028 -0.012  0.004  0.005 0.013 -0.018
3 0.028 -0.016  0.005 -0.012 0.026 -0.021
4 0.027 -0.025 -0.004 -0.008 0.026 -0.020
5 0.022 -0.021 -0.017 -0.006 0.019 -0.012
6 0.036 -0.019 -0.003  0.026 0.011  0.001
> head(pdf)
  Predictions     x1    x2     x3     x4     x5    x6     x7     x8    x9
1 Prediction1 -0.009 0.028  0.007 -0.009 -0.063 0.023  0.020 -0.022 0.000
2 Prediction2 -0.016 0.068 -0.005  0.011 -0.068 0.043  0.017 -0.036 0.007
3 Prediction3 -0.014 0.059 -0.017  0.045 -0.052 0.090  0.009 -0.047 0.021
4 Prediction4 -0.029 0.042 -0.029  0.050 -0.046 0.121 -0.019 -0.018 0.028
5 Prediction5 -0.038 0.032 -0.037  0.079 -0.024 0.130 -0.005 -0.026 0.031
6 Prediction6 -0.062 0.058 -0.027  0.087  0.022 0.124 -0.016 -0.036 0.047
     x10    x11    x12    x13    x14    x15
1 -0.027 -0.037  0.007  0.012  0.023 -0.026
2 -0.061 -0.029 -0.010  0.000  0.048 -0.027
3 -0.073 -0.035 -0.004 -0.003  0.048 -0.023
4 -0.045 -0.041  0.000 -0.001  0.048 -0.025
5 -0.034 -0.024 -0.038  0.037  0.030 -0.007
6  0.005 -0.020 -0.045  0.064 -0.002 -0.005

Finally yielding this plot:

And this faceted ecdf plot: