From: Doug Jones (djones@xcor.com)
Date: Sun May 13 2001 - 13:26:27 MDT
Adrian Tymes wrote:
>
> Spike Jones wrote:
>
> [re: ExtroQuiz]
>
> > Guys we need to set up some test cases and run em
> > when you are here for e5. spike
>
> For this we have simulations. This is a pretty simple case...
>
> Case A: Host says nothing, so contestant never has a reason to switch.
>
> Case B: Host reveals a nonprize door after initial choice, always
> choosing the contestant's door if the contestant chose a nonprize door;
> contestant switches only if contestant's door is revealed.
>
> Case C: Host reveals a nonprize door completely at random; contestant
> only switches if contestant's door is revealed.
>
> Case D: Host reveals a nonprize door completely at random; contestant
> always switches.
>
> Each trial was for 1000 runs; the number given is the number of times
> the contestant won in that trial of that case.
> Trial 1 2 3 4 5 Total WinRatio ExpectedWinRatio
> CaseA 323 302 360 322 331 1638 0.3276 0.3333
> CaseB 674 670 675 661 662 3342 0.6684 0.6666
> CaseC 536 534 478 515 515 2578 0.5156 0.5000
> CaseD 519 493 489 506 477 2484 0.4968 0.5000
>
> Any other strategies I should run? I've included the code (in Perl)
> below, if anyone else would like to double-check my work.
Case E: Host reveals a nonprize door only if the contestant's first
choice is the prize. Contestant always switches when given a choice.
WinRatio of zero, if contestant has no learning function, rising to one
if contestant figures it out and host doesn't change strategy- which of
course he will. I'm amazed nobody ever punched Monty's lights out.
Doug Jones, Rocket Plumber
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:07:38 MST