Build data frame from multiple CSVs in R - r

Question

(:
I don't know much about R, but I was required to plot a set of box plots from a data ensamble using it.
I have a set of .csv files representing a set of 2D data. They contain the following columns:
i: the row of the matrix
j: the column of the matrix
VBoot: a property of the matrix
My data is 128 x 128, but the .csv just contain indices for non-zero properties.
I have to plot a box plot for each of these files, side by side.
This is my approach:
library(ggplot2)
library(reshape)
# Set the directory to read the files
setwd("/Users/me/data/CSV/")
operatorProperty <- function(operator, property, degrees, m, n)
{
p <- list()
for (degree in degrees)
{
file <- paste(c(degree, operator, property, ".csv"), collapse="")
data <- read.csv(file, header=TRUE, sep=" ", dec=".")
# Create an array m * n to fill with the data
b <- vector(mode="double", length=(m*n))
# Rebuild the complete data to properly build the box plot
b[data$i * m + data$j] = sqrt(data$VBoot)
p <- append(p, list(b))
}
p
}
So far, I just created a list to insert the data for each ensamble.
Then, I though I should build a data.frame:
min_degree = 0
max_degree = 45
delta = 5
m = 128
n = 128
degrees <- seq(min_degree, max_degree, delta)
property <- "VBoot"
operator <- "Prewitt"
Sobel <- operatorProperty(operator, property, degrees, m, n)
df <- data.frame(degrees, Sobel)
df2<- melt(data=df,id.vars="degrees")
p <- ggplot(df2, aes(x=degrees,y=value,colour=variable)) +
geom_boxplot() +
theme(legend.title=element_blank()) +
xlab(expression(theta)) +
ylab("Bootstrap Variance")
However, I can't build the data.frame. I don't know how to proceed. An example of the data can be found here.
Thank you in advance.

Answer 1

Ok. Well, I had to change a few things to get this to work with the sample data. Here's the setup
m = 128
n = 128
operatorProperty <- function(operator, property, degrees, m, n)
{
Map(function(degree) {
file <- paste(c(degree, operator, property, ".csv"), collapse="")
data <- read.table(file, header=TRUE, dec=".")
# Create an array m * n to fill with the data
b <- vector(mode="double", length=(m*n))
# Rebuild the complete data to properly build the box plot
b[data$i * m + data$j] = sqrt(data[[property]])
b
}, degrees)
}
degrees <- c('00','05')
property <- "MSE"
operator <- "Prewitt"
Sobel <- operatorProperty(operator, property, degrees, m, n)
With this modified form, Sobel is a list with named elements corresponding to the different degrees. We can turn this into a data.frame and plot with
df2<- melt(data.frame(Sobel, check.names=F))
p <- ggplot(df2, aes(x=variable,y=value,colour=variable)) +
geom_boxplot() +
theme(legend.title=element_blank()) +
xlab(expression(theta)) +
ylab("Bootstrap Variance")
which looks very funny because you have so many zeros. All of your non-zero entries are just marked as outliers.
But even if we didn't nicely name the Sobel list, It basically was a list with two vectors (once for each degree)
list(c(0,0,0,0, ...), c(0,0,0,0,...))
if you wanted to merge that with degrees and turn into a data.frame, another choice could have been
do.call(rbind, Map(cbind.data.frame, degrees, Sobel))