Whereas, what immediately pops into my mind is that you should use a database ... for example, “SQLite,” which is a universally-available database system that uses simple files – no database server.   While you might (or might not) find yourself having to go through some calisthenics to get the data into a proper database format, once you have done so it appears to me that many of your subsequent activities can readily be accomplished using database JOINs and summary features such as GROUP BY.   Furthermore, once you do get your data into that form, many tools – including any ubiquitous “spreadsheet” – might be put to the task.   (Data volumes and relationships such as you describe are “thoroughly un-exciting” to these database technologies.   They do this sort of thing for breakfast.   “A million rows?”   Yawn ...)

In your grander investigations, I would also point you to the novel open-source statistical programming language, “R.”   It can do a great many useful things in ways that are a breath of fresh air versus other packages such as SPSS® and SAS®.

So, if it seems that I seem to be nudging you away from Perl, actually I am not.   In my opinion, Perl is (still) the locomotive-of-choice for handling data files.   But, do not automatically jump to the conclusion that you will find yourself writing custom programs to do this entire job, because quite-likely you won’t have to.