yes, but you have to do a coalesce(1). This will generate a single csv file, however you will also lose some parallelism as this coalesce(1) is propagated upstream.

Also do not forget to disable the writing of _SUCCESS etc files (see this topic)