This approach provides a simple way to compare the 'attractiveness' of different candidate projects.

 These criteria are not generic. You need to consider what is relevant for your company - but hopefully they provide an idea. Each individual attribute has it's standalone own weighting (importance - shown as a to k). These attributes are grouped and each group has it's own weighting that is applied to all the individual attributes to produce a composite percentage.

The scores for each project by attribute are then multiplied by these composite percentages to provide a score per attribute and totalled to give an overall score per project.

The pragmatic way to work with the attribute and group weightings is to produce a draft set of weightings and then start to put projects through the process. The horrified looks around the table will show you when the projects being selected by this approach are intuitively the wrong ones. At that point you need to re-work the weightings until you start picking the right ones. The best projects to pick to tune this approach are probably ones that have become successful innovations/products/features.

This is one end of the scale of complexity.  I'll get round to writing about efficient frontiers etc. another time.