[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Explaining (J)SD w/GNU & rollouts

Posted By: Matt Cohn-Geier
Date: Tuesday, 16 October 2007, at 6:39 a.m.

In Response To: Explaining (J)SD w/GNU & rollouts (Bryce)

This seems like a good overview of standard deviations and error and the like to me. I'm hardly a statistician, so I can't really claim whether it's technically accurate or not, but it looks pretty good to me.

One phrase that always rubs me the wrong way:

There is a 68% chance that A lies in the interval (.0773-1*.0043, .0773+1*.0043) = (.0730, .0816)

A has a definite value. Therefore whether A lies on the interval (x,y) is either true or false, and has a probability of 1 or 0. To say there is a 68% chance that A lies on an interval is like saying there is a 68% chance that 0 is not 1.

What there is a 68% chance of is that the procedure we use (in this case, a GNU rollout) will generate a confidence interval that contains A. IOW, if we were to run 100 rollouts with different seeds, we would expect 68 of the intervals to contain A, and 32 to exclude A.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.