R plotting dataframe and keeping track of factor to number conversion - r

Question

He is my dataframe and its plot
my_df <- data.frame(var_1= as.factor(sample(c(0,1), 10, replace = TRUE)),
var_2 = sample(1:20, 10, replace = TRUE),
var_3 = as.factor(sample(c('a','b', 'c'), 10, replace = TRUE)))
plot(my_df)
So factors are getting converted to numeric values. How can I figure out map between factor value and its numeric representation? For example it looks like var_3 have the following conversion is in place a -> 1, b -> 2, c -> 3.
Also can I display this conversion map on the graph?

Answer 1

plot.data.frame passes ... arguments to ?pairs which has a labels argument
my_df <- data.frame(var_1= as.factor(sample(c(0,1), 10, replace = TRUE)),
var_2 = sample(1:20, 10, replace = TRUE),
var_3 = as.factor(sample(c('a','b', 'c'), 10, replace = TRUE)))
plot(my_df, labels = LETTERS[1:3])
So just get a vector of labels and use that
f <- function(data, default = names(data), use.varname = TRUE) {
default <- rep_len(default, ncol(data))
sapply(seq_along(data), function(ii) {
x <- data[, ii]
if (is.factor(x)) {
lbl <- paste(levels(x), seq.int(nlevels(x)), sep = ' -> ', collapse = '\n')
if (use.varname)
paste(default[ii], lbl, sep = '\n') else lbl
} else default[ii]
})
}
f(my_df, use.varname = FALSE)
# [1] "0 -> 1\n1 -> 2" "var_2" "a -> 1\nb -> 2\nc -> 3"
f(my_df, use.varname = TRUE)
# [1] "var_1\n0 -> 1\n1 -> 2" "var_2" "var_3\na -> 1\nb -> 2\nc -> 3"
plot(my_df, labels = f(my_df))

Answer 2

A better option might be to use ggpairs that gives more information
library(GGally)
ggpairs(my_df)

Answer 3

The integers corresponds to the levels of your factor. If you want a scatterplot as that reported in your example, you can simply set the labels for your axis using the axis function.
plot(as.numeric(my_df$var_3), as.numeric(as.vector(my_df$var_1)), axes = F)
axis(side = 1, labels = levels(my_df$var_3), at = 1:length(levels(my_df$var_3)))
axis(side = 2)
box()
Now, if you want a multi-plot result, you can do as follows.
par(mfrow=c(3,3))
for (i in 1:ncol(my_df)){
for (j in 1:ncol(my_df)){
if (i == j) {
plot(1, cex = 0, ylim = c(0,2), xlim = c(0,2))
text(1, 1, labels = paste(names(my_df)[j]))
} else {
plot(as.numeric(my_df[,i]), as.numeric(my_df[,j]), axes = F,
xlab = names(my_df)[i], ylab = names(my_df)[j])
if (is.factor(my_df[,i])){
axis(side = 1, labels = levels(my_df[,i]), at = 1:length(levels(my_df[,i])))
} else {
axis(side = 1)
}
if (is.factor(my_df[,j])){
axis(side = 2, labels = levels(my_df[,j]), at = 1:length(levels(my_df[,j])))
} else {
axis(side = 2)
}
box()
}
}
}
par(mfrow=c(1,1))
Definitely verbose and not very pretty, but as you can see the variable levels (class names) are retained and plotted at the axis ticks...