n-Dimensional Array: Optimizing Eliminations

[bsmither]

I've been searching for a science that will solve this analysis problem. Since I don't know what I'm looking for, I cannot properly express the task to accomplish.

 

I have an eCommerce store selling a product that has (let's keep it simple) three properties and each property can hold any one of three possible relevant values (or null).

 

Combinations: 3^3=27

 

A typical combination would be round/solid/blue, or cubic/squishy/white.

 

The task is to efficiently eliminate that part of the array if any particular {set of values} is null. That is, if I currently have no pink items, the relevant sections of the array would get eliminated.

 

Product variations with low n-values is easy to manually handle. But with five and above, that's the math(?) theory I am trying to know the name of.

 

(I wonder if any Matlab manuals would give a clue.)

 

I am trying this elimination of sections of an established array instead of rebuilding the array every time a value is locked in.

 

On the other hand, maybe I'm trying to be too smart about this and should just let the server do the grunt work over and over.

[requinix]

I'm wondering if you've given a complicated description of the following:

shapes   = { round, cubic, cylindrical }
textures = { solid, squishy }
colors   = { red, orange, yellow, green, blue, purple, black, white }

items = {
  { round, solid, blue },
  { cubic, squishy, white },
  { round, solid, orange },
  { round, solid, white }
}

unused shapes   = { round, cubic, cylindrical } - { round, cubic, round, round } = { cylindrical }
unused textures = { solid, squishy } - { solid, squishy, solid, solid } = { }
unused colors   = { red, orange, yellow, green, blue, purple, black, white } - { blue, white, orange, white } = { red, yellow, green, purple, black }

That's just a 2D array (1D is property, 2D is value) and you can get the "unused" values with array_diff. Edited October 23, 2014 by requinix