"Unique" Column w/ a 1-to-Many/Many-to-1 Relationship & "Child Rows" + Math on Grouped Rows

tylerlweb

We're using MongoDB with a pretty large dataset (about 3 million records). We're using PHP to reach out to Mongo & spit out everything DT needs. We've got a number of things working nicely - pagination, drag & drop of columns w/ live preview, multi-column sorting, and a custom interface to build the $criteria & $projection for Mongo. Very glad to pay $9, I wouldn't mind paying more for this incredible tool! If you need an extra donation to handle this question I'll understand.

We need to add 2 more big features:
1. We need to be able to select just one column and have that be the chosen "unique" column, so that if 30 of the response records have the same value in the chosen column, we only want one row for those 30 records. But we need all 30 rows worth of data to be thrown into that one row (kinda the same idea as an HTML rowspan).
2. This is scientific data, so we also need to be able to get the arithmetic mean (and hopefully the geometric mean) of the numeric columns for the rows that are grouped together, so take the chosen column, look at those 30 inline "fake" cells, and turn them into 1 number.

Of course, the 30 records might actually be 1,200 records, who knows - so it's important that there be virtually no limit on the number of results shown inside the cells (and absolutely no limit on the number of records we're doing the math on). Hoping there's some extension/plugin that I'm missing - if you could show me an example of something like this or just get us going in the right direction that'd be amazing.

Thanks!

EDIT: BTW, I noticed this ColReorder FixedHeader page isn't working right - it errors out w/ "Undefined is not a function" http://datatables.net/release-datatables/extensions/ColReorder/examples/fixedheader.html