This is certainly not a common task, hence, no comments or answers. However, a good starting point for you would be to build upon existing packages that can determine copy number from the 450K methylation array:
Once you identify global profiles of copy number, you can apply some extra customised steps to estimate ploidy. I did this previously (unpublished) but based on NGS depth of coverage profiles. It was surprisingly precise.
Edit: In fact, there appears to be a possible logical next step: If you derive global copy number segment profiles using one of the above-mentioned programs, then you could feed these into ABSOLUTE in order to detect ploidy.