/sci/ - Science & Math » Thread #12861966

610KiB, 1440x2086, jke.jpg

View Same Google iqdb SauceNAO

Anonymous Sun 21 Mar 07:23:12 2021 No.12861966 View Reply Original Report

Quoted By: >>12862005 >>12862185 >>12862194 >>12862197

Is n=30 always sufficient?

Anonymous

Anonymous Sun 21 Mar 2021 07:44:50 No.12862005 Report

Quoted By:

>>12861966
no. Let's say you have a rare event that occurs randomly once in a million years and you can only run a simulation a year long. In just 30 simulations you may never capture the event

Anonymous

Anonymous Sun 21 Mar 2021 09:21:30 No.12862185 Report

Quoted By:

>>12861966
No. Also 30 is kinda arbitrary.
For stable distributions it can take much longer to converge.
There is also a dependence on the dimension of your random variable.
https://en.wikipedia.org/wiki/Berry%E2%80%93Esseen_theorem

Anonymous

Anonymous Sun 21 Mar 2021 09:25:04 No.12862194 Report

Quoted By:

>>12861966
No. Bigger number better number

Anonymous

Anonymous Sun 21 Mar 2021 09:26:57 No.12862197 Report

Quoted By:

>>12861966
it's a rule of thumb that's easy to remember... but personally I like studies with n > 100 as a standard

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches