11-14-2023, 04:35 PM
(11-14-2023, 01:34 PM)RhoSigma Wrote:(11-14-2023, 10:35 AM)SMcNeill Wrote:(11-14-2023, 09:54 AM)bplus Wrote: This is crazy to order the columns most simialr to all the other columns???
You can only compare all the other columns to one of the columns for a proper compare of matches.
Otherwise a column thats all *'s will match perfectly any other column with random *'s if that's what you really want then order the columns by most *'s
Not really, I don't think. Take a look at this example:
x----
x----
x----
x----
-xxxx
Now the first column has 4 stars, but it doesn't match any of the other columns.
The other columns only have 1 star, but they all match in the same row.
I'm thinking that it might be the 4 columns that we're looking for here, rather than the 1 column with all the stars?
I not really followed this, but it seems like a normal binary number sort to me, instead of *--**-**, think of it as 10011011. That's a number and can be perfectly sorted with other numbers (similar bit patterns).
I don't think it's quite that simple either, Rho. Think of this set of data:
x----
x----
x----
x----
-xxxx
---xx
Now, if you just look at going down, column 1 has the most matches.
If you just look going right, row 4 has the most matches.
I think what he's looking for here is columns 4 and 5, which each have 2 stars matching between them.
The only way I see to get this is basically just as bplus said: "You can only compare all the other columns to one of the columns for a proper compare of matches."
The trick is you just have to do a loop to compare ALL of those columns 1 by 1, against all of the other columns, and then shuffle the results of those matches.