CODEWITHSUNDEEP
rlattice
saffsd
saffsd

Reputation: 24312

How do I produce a barchart of one factor conditioned on another factor?

Given a data frame like the following, with 2 or more factors:

 data <- data.frame(V1=sample(c('A','B','C'), 50, T), V2=sample(c('X','Y'),50,T))

I can produce a barchart of the tabulation of a factor independently, for example:

library(lattice)
with(data, barchart(V1))

How can I produce a multi-panel plot conditioned on the value of the second factor? The syntax I would expect to use is:

with(data, barchart(V1 | V2))

But this produces the following warning:

Warning message:
In Ops.factor(V1, V2) : | not meaningful for factors

and no meaningful output.

Upvotes: 3

Views: 189

Answers (1)

mnel
mnel

Reputation: 115425

When you call with(data, barchart(V1)) you are calling barchart on an object of class factor and it is calling barchart.default, which calls the equivalent of barchart(table(V1)) (and thus calls barchart.table) producing the plot you want.

If you want to have both V1 and V2 included you have to use them in your table, ie. 

with(data, barchart(table(V1, V2))

enter image description here

You can then use groups = TRUE to set the last dimension as the grouping variable

eg

with(data, barchart(table(V1, V2), groups = TRUE))

enter image description here

If you want to use the formula method, you will have to calculate the values yourself beforehand.

eg

 library(data.table)
 DT <- data.table(data)

 barchart(V1~N|V2,DT[, .N,by= list(V1,V2)])

which gives the same result as with(data, barchart(table(V1, V2), groups = TRUE))

Upvotes: 7

Related Questions