3 ms·
Oh, and another thing, so I'm making another comment. > Input Elimination: Often by using - we find that a column is unnecessary I've found this to be true in
by dragontamer 6y ago
Oh, and another thing, so I'm making another comment.
> Input Elimination: Often by using - we find that a column is unnecessary
I've found this to be true in many cases. However, efficiently calculating a potential "input elimination" has eluded me.
I've done research into various relational-database techniques. I've looked at binary decision diagrams (or their relational multi-valued counterparts: Multi-Valued Decision Diagrams). Etc. etc.
It is "obvious" that minimizing columns and removing unnecessary columns can grossly improve performance. But I can't find any efficient algorithm that detects this case.
Lets say we have Table ABCD (with 4 columns: ABCD). To test if Column C can be eliminated, I came up with:
ABD = (Select a, b, d from TableABCD where C = true)
ABD2 = (Select a, b, d from TableABCD where C = false)
If ABD == ABD2, then C is redundant.
That's for binary variables. If C has multiple values, then you have to do this for every value C could have.
-------
This seems inefficient to me. Especially if I'm checking every column to see if they're redundant. (Repeat the above process for A, B, and D). I feel like there has to be some kind of dynamic-programming magic (or even Greedy-algorithm) that can figure out the redundant columns all simultaneously. I just don't feel like coming up with the algorithm.
I can't believe that I'm the first one to investigate this problem. But I don't know the "magic word" for this to search on. I don't even know how to search the internet for this algorithm, even though I know its important. If anyone knows the most efficient solution, please let me know :-)