User:Cook Me Plox/trekkies
Open questions about Temple Trekking data.
- 1 What's the relationship between location (#1958) and event (#1957)?
- 2 What's the distribution of events?
- 3 Can you get the same event back to back?
- 4 Can you get the same event in a trek twice, even NOT back to back?
- 5 If you escape/teleport out, can you get the same event immediately after you escaped?
- 6 Can you get sea monster back to back on 17 and 18?
- 7 What's the set of locations you can get on each trek, going in either direction?
- 8 What's the distribution of locations?
What's the relationship between location (#1958) and event (#1957)?[edit | edit source]
- Seems that location 1 (Mort'ton) is always shades (event 1)
- Locations 17 is always a sea monster when going from Burgh de Rott to Paterdomus, and location 18 is always a sea monster when going from Paterdomus to Burgh de Rott. When encountering those locations in the other direction, they are a normal roll (see below).
- Every other location appears to follow the same distribution of events 3 to 12, although we don't quite want to use the sample mean because of repeats.
What's the distribution of events?[edit | edit source]
- Best guess is the following:
- 3 (juvinates): 20/128
- 4 (ghasts): 18/128
- 5 (snails): 13/128
- 6 (snakes): 13/128
- 7 (bog): 15/128
- 8 (trees): 12/128
- 9 (rope): 7/128
- 10 (Abidor Crank): 4/128
- 11 (lumberjacks): 14/128
- 12 (nail beasts): 12/128
This is based on about 12,000 individual events. I've attempted to handle the "no repeated events" policy by doing a sort of Jacobi-iteration-inspired algorithm to guess how many of the duplicate events were essentially avoided, which I then threw into our log-likelihood denominator guesser, giving this plot:
There's a hump at 128 that is fairly noticeable, but we also don't have enough data to answer definitively – we probably need about 4-10x as much.
Can you get the same event back to back?[edit | edit source]
- Seems like NO in most cases, even across separate treks.
- Hypothesis: the random roll for events 3 to 12 won't let you get the same event as what is already set on varbit #1957.
- This has no effect on shades and sea monsters, which are not part of the random roll. There's many cases of getting shades as the last event of one trek and the first event of the next, and similarly for sea monsters.
Can you get the same event in a trek twice, even NOT back to back?[edit | edit source]
- YES. Limiting the dataset to cases of complete treks that have exactly 3 events and don't have shades or sea monster, we see 128 total events. 15 of these have the same event in slot 1 as slot 3, which is close to the expected true distribution (around 12%-13%).
- Similar things hold if you look at 4-event treks.
- This is surprising, but I think people rarely do 3+ event treks.
If you escape/teleport out, can you get the same event immediately after you escaped?[edit | edit source]
- YES. 418 cases of (nonzero event, zero event, 1956 set to 1, nonzero event). In 49 of these, the same event was at the beginning and end, again close to the expected 12%-13%.
Can you get sea monster back to back on 17 and 18?[edit | edit source]
- NO. From further testing, the location you get a sea monster on (17 vs 18) is dependent on which direction you are going. It's therefore impossible to get sea monsters in both.
What's the set of locations you can get on each trek, going in either direction?[edit | edit source]
- It appears that within the basic confines of which locations can be reached in a particular route difficulty, every location can lead directly to every other location.
What's the distribution of locations?[edit | edit source]