first of all i would like to thank dinglemouse for this really awesome kata, even though i haven't solved it yet, i have had fun pondering it so far. the simple tests have all been successful, but in the advanced tests, 2 fail: fire_drill and random. When I read your question, I wondered how you got the test case of fire_drill, for example?

calculation of the averages is wrong, isnt it?

actually the kata was fun, but due to the inaccurate descriptions and the many assumptions you have to make, it is unsatisfying. It is strange when you "must" write "wrong" code...

i decided to cancel the kata to save myself from headaches.