Add standardized mob count as a weighted element in map score calculation

[kuarcis]

The original data on mob count is based on 10 runs of a map. The author has also included the difference between the maximum and minimum mob count in each map.

By standardizing the average mob count using the formula (count_average^2)/(count_max-count_min), we can expect to generate a rank that promotes maps with less volatile mob counts and higher mob counts. This is achieved through two parts of the formula: 1) count_average/(count_max-count_min) promotes maps with less volatile mob counts, and 2) count_average*count_average/(count_max-count_min) promotes maps with higher mob counts.

This standardization process and result can be treated as another weighted element. Users can set the weight to 0 to get the same result as before or make it non-zero to use the standardized mob count result.

[deathbeam]

Does map mob count being volatile or not even matter in the long run? Currently what I do for scoring the maps is:

ceil(10 * (mob_count - min(mob_counts)) / (2ndmax(mob_counts) - min(mob_counts)))

I do 2nd max for mobs counts when getting max because desert spring is big outlier for mob counts and a lot of mobs there are in boss room.

This assigns maps score from 0 to 10 based on the average mob count relative to other mob counts. I think thats good enough because in long run only average matters anyway and mob count also isnt the only metric or even the most important one when choosing maps anyway.

[kuarcis]

I came up with the idea that some maps seem to have more volatile monster counts than others. For example, the Crate map in the source counting file is the fifth highest mob count map, but it also has a huge range of mobs (the second largest range to be precise). In the long run, this difference will appear in the total mob count level. If we count by average killing each run in the same map, both method would show quite similar suggestion if all maps show similar volatility, as you mentioned. This is just another way to measure “how many mob kill counts can I expect?”

this method can also deal with the desert spring outlier issue on a certain level, making desert spring looks less attractive.