Question

I have a table like follows:

Potion { id, name, freq }

I gave away potions to users in ‘random’ manner. but I want to be able to control frequency of the randomness because some potions is more powerful than others. What’s the best way to do this?

I am thinking of something like this:

  id | name  |   freq
  -------------------
   1 | light |     5
   2 | fire  |    10
   3 | water |    10
   4 | earth |    10
   5 | air   |     5

light and air are more powerful than fire, water, and earth so they have smaller frequencies.

sum = select sum(*) from potions;
x = random(1..sum)
table = select * from potions;
offset = 0
for each element in table
  if offset + element[freq] > x
     chosen = element[name]
  end
  offset += element[freq]
end

I think the code above (ruby + sql) will works but its not scalable and i very much doubt it that it’s the most efficient way of doing it. Could anyone please help me to do so?

Thank you in advance,

Answer 1

You first need to create a view on top of RAND() so it returns the same value for each invocation during execution (otherwise you’ll get weird results):

create view constant_rand as select rand() as rand;

Next select a random row weighted by the cumulative freq:

select id
from potion p1
join constant_rand
where rand * (select sum(freq) from potion) between
    -- sum of freq below this row
    ifnull((select sum(freq) from potion p2 where p2.id < p1.id), 0)
    -- sum of freq below and including this row
    and (select sum(freq) from potion p2 where p2.id <= p1.id);