I have a database full of units (U1 - U50, U51...) where every unit has the same standard attributes (A1 - A10) and where a % of each attribute defines the amount of that attribute for that particular unit.
Here is a Dropbox link to the Excel document, for reference.
Defining a Unit by it's value of the standard Attributes looks something like this:
-
With that, there is a relationship that each Attribute has with every other. Some attributes are similar to another attribute, while others are not.
The Attribute-to-Attribute similarity matrix looks something like this:
-
Lastly, a User has the ability to rate any random Unit, at which point it gets stored in a special list.
The User-Rated list looks something like this:
-
The goal is to produce a separate list of the units that are the most similar to each unit in the user-rated list, while taking into account what rating the user has given to all the other units in their list.
Essentially: Upon rating a new unit, what units from the unit-database would the user like the most (given their history of other unit ratings)?
The algorithms (and Excel formulas) needed for this are beyond my knowledge. I've set this up in Excel the best way I know how. There might be better ways of solving this by altering the data (i.e. making the Attribute-similarity matrix a collinear matrix). I just don't know. To anybody who can solve this, thank you. If you can perhaps share with me what proper mathematical areas are included in this sort of problem, that would be fantastic. If you can work out how to do this in Excel, that would be amazing.