Services & ResourcesWolfram ForumsMathGroup Archive

[Date Index] [Thread Index] [Author Index]

Re: Quantile and InverseCDF

• To: mathgroup at smc.vnet.net
• Subject: [mg87875] Re: [mg87819] Quantile and InverseCDF
• From: Darren Glosemeyer <darreng at wolfram.com>
• Date: Sat, 19 Apr 2008 03:33:57 -0400 (EDT)
• References: <200804180637.CAA12464@smc.vnet.net>

Peter Breitfeld wrote:
> (* Version 6.0.2 for Mac PPC *)
>
> I thick I am missing something obvious:
>
> The documentation for InverseCDF says:
>
> "For a discrete Distribution dist the inverse CDF at q ist the largest
> integer x such CDF[dist,x]<=q.
>
> But I get
>
> binom=BinomialDistribution[20,0.3]
>
> InverseCDF[binom,0.5]     ---> 6
> Quantile[binom,0.6]       ---> 6
>
> CDF[binom,7]              ---> 0.772...
> CDF[binom,6]              ---> 0.608...
> CDF[binom,5]              ---> 0.416...
>
> so the quantile ist the smallest integer such CDF[dist,x]>=q  ??
>
>
> Gruss Peter
>   

Thanks for pointing this out so we can correct the documentation. You 
are right that the statement should be that it is the smallest integer 
such that CDF[dist, x]>=q.

Darren Glosemeyer
Wolfram Research

• References:
  • Quantile and InverseCDF
    • From: Peter Breitfeld <phbrf@t-online.de>