At Metabase we're using kixi.stats quite extensively and started to run into performance issues. This PR fixes the low hanging fruit type hints and results in speedups between 50% and 1000%.
One potential issue: I cast everything to double. This could be considered a regression in the sense that none of the functions return a precise (long) result anymore. On the plus side, this also means things stop exploding with overflows on large inputs.
At Metabase we're using kixi.stats quite extensively and started to run into performance issues. This PR fixes the low hanging fruit type hints and results in speedups between 50% and 1000%.
One potential issue: I cast everything to double. This could be considered a regression in the sense that none of the functions return a precise (long) result anymore. On the plus side, this also means things stop exploding with overflows on large inputs.