User:Flargen: Difference between revisions
From A KoL Wiki
imported>Flargen No edit summary |
imported>Flargen mNo edit summary |
||
| (20 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
I decided to have a blue name instead of a red one. A somewhat useful piece of information: if you ever see a Flargen in the game then that's not me. Unless it turns out no one has the name and I get a wild hair up my hindquarters and decide to get a new multi. Such a sequence of events has yet to occur, however. I am otherwise simply a more-or-less typical | I decided to have a blue name instead of a red one. A somewhat useful piece of information: if you ever see a Flargen in the game then that's not me. Unless it turns out no one has the name and I get a wild hair up my hindquarters and decide to get a new multi. Such a sequence of events has yet to occur, however. I am otherwise simply a more-or-less typical hardcore speed ascender who sometimes moonlights as a snarky wiki admin and spader. | ||
==Spading Projects== | |||
'''Current''': | |||
*Level 1 zone drop rates | |||
'''Future''': | |||
:To be determined | |||
== | ===Item Drop Rates=== | ||
I mostly specialize in working out the base drop rates of the various items dropped by the kingdom's many monsters. | |||
*I (generally) collect data until there is a unique integer in the 95% CI obtained from the observations. | |||
**[http://sweb.uky.edu/~tjmart1/kol/bayes.html Starwed's] tool can be used to obtain exact calculations on the predictive capacity of integer drop rates. This converges at about the same time as the above, and sometimes faster, but it's a bit more onerous for me to input the data properly into that then it is to do it the aforementioned way, especially when you want results for several items and several monsters at once. | |||
*Data posted before 9/20/11 was analyzed as a simple sum of (scaled) bernoulli trials, with no weighting (other than the scaling, which comes by dividing out the drop rate multiplier). Data after that is analyzed by weighting the information from each drop bonus by inverse variation. This will make Yiab's high-error, low bonus spading more reusable. The previous method would overemphasize the sheer number of encounters Yiab had done, resulting in less conclusive results than if Yiab's data was excluded. | |||
Latest revision as of 12:47, 3 December 2011
I decided to have a blue name instead of a red one. A somewhat useful piece of information: if you ever see a Flargen in the game then that's not me. Unless it turns out no one has the name and I get a wild hair up my hindquarters and decide to get a new multi. Such a sequence of events has yet to occur, however. I am otherwise simply a more-or-less typical hardcore speed ascender who sometimes moonlights as a snarky wiki admin and spader.
Spading Projects
Current:
- Level 1 zone drop rates
Future:
- To be determined
Item Drop Rates
I mostly specialize in working out the base drop rates of the various items dropped by the kingdom's many monsters.
- I (generally) collect data until there is a unique integer in the 95% CI obtained from the observations.
- Starwed's tool can be used to obtain exact calculations on the predictive capacity of integer drop rates. This converges at about the same time as the above, and sometimes faster, but it's a bit more onerous for me to input the data properly into that then it is to do it the aforementioned way, especially when you want results for several items and several monsters at once.
- Data posted before 9/20/11 was analyzed as a simple sum of (scaled) bernoulli trials, with no weighting (other than the scaling, which comes by dividing out the drop rate multiplier). Data after that is analyzed by weighting the information from each drop bonus by inverse variation. This will make Yiab's high-error, low bonus spading more reusable. The previous method would overemphasize the sheer number of encounters Yiab had done, resulting in less conclusive results than if Yiab's data was excluded.
