As Kostas said, create a *new* group variable according to your needs:

plot(Petal.Width~Species,iris)

iris$Species2

From boxplot manual page:

boxplot(formula, data = NULL, ..., subset, na.action = NULL)

where

formula: a formula, such as y ~ grp, where y is a numeric vector of data values to be split into groups according to the grouping variable grp (usually a factor).

So, create the grouping variable grp variable according to your needs

Thanks kostas, but this is exactly my problem. The definition of the order of the grouping variable just doesn't seem to ever be right. Here's an example:

Say you have two columns v1 with numeric variables, v2 with categorical (4 groups, grpA, grpB, grpC, grpD). The easy boxplot formula would be boxplot(v1~v2). But you don't won't grpA to be the first boxplot, but the boxplot to read D -> A -> B -> C.

So how do you then "redefine" your column? grp

As Kostas said, create a *new* group variable according to your needs:

plot(Petal.Width~Species,iris)

iris$Species2

Is this what you mean?

http://stackoverflow.com/questions/4260698/r-ordering-in-boxplot

I agree with Eik,

and here is another quick example

########### R-code ###############

dats2

Hi guys, thanks so much for your replies.

So Gregory directed me to the SO page which I think made me consolidate what Eik and Kostas wrote. Thanks!

I personally find the follwing the most simple (using the same example as above v1 with numeric variables, v2 with categorical (grpA, grpB, grpC, grpD) - which is just a simplification of your code Kostas & Eik:

mydata$v3

Thanks for that!

Cheers,

O.

Hey guys! There is an exemple that illustrates all the good explanations above in the R graph gallery.

http://www.r-graph-gallery.com/22-order-boxplot-labels-by-names/

Thanks oliver: This script ; mydata$v3

I have tried all the possibilities suggested along the thread and the only one who worked for me was the one of Patrick Kimani. Thanks everybody to post theirs solutions but thank YOU Patrick ¡¡¡ yours was the simplest script which matched perfectly with my basic knowledge in R and interestingly the only one who paid off ¡¡¡ Thanks again you all ;D

I don't know what to do!

I tried all of them but the problem is , because I want to plot only a subset of my entire data frame, say I have a data frame with 3 columns a, x and y and I want to plot only a subset group of the entire data: boxplot (x~y, data=df[a=="..."]).

when I try to use the factor function to re-order the levels of a, and i assign this function to a vector; "ordered_a", then I try to do the same boxplot I mention above, it doesn't work.

there's a message saying

#Error in stats::model.frame.default(formula = x ~ y :

variable lengths differ(for ordered_a)

IF anyone has crossed the same problem please let me know how you figured it out or where else should I be looking for answers!

Thank you very much!