in reply to Levenshtein distance: calculating similarity of strings

This code uses a hash instead of a simple array for mat. I changed it to a regular array, and it reduced the execution speed by 19.3%.
  • Comment on Re: Levenshtein distance: calculating similarity of strings

Replies are listed 'Best First'.
Re^2: Levenshtein distance: calculating similarity of strings
by AnomalousMonk (Bishop) on Feb 21, 2018 at 22:31 UTC
    ... it reduced the execution speed ...

    So just to be clear, are you saying your change made the code slower by about 20%?


    Give a man a fish:  <%-{-{-{-<

      Oops, sorry. It reduced execution time by about 20%. I'm not allowed to post my actual code, but it was just the few obvious changes: declare @mat rather than %mat, and use [] rather than {}.

      I made a loop that called levenshtein with 2 arrays of strings of 2-character words. Between one and 25 words per string, comparing all possible combinations of 51 different such "sentences". Hence 51 * 51 = 2061 string comparisons of between one and 25 words per sentence. I executed it on an old MacBook Pro 200 times, and took the average "user" time from the "time" command.

      Average "user" time was 19.31% shorter.

        There's also Benchmark.


        Give a man a fish:  <%-{-{-{-<