Re: Re: Counting Runs
- To: mathgroup at smc.vnet.net
- Subject: [mg52032] Re: [mg52018] Re: Counting Runs
- From: DrBob <drbob at bigfoot.com>
- Date: Mon, 8 Nov 2004 03:13:11 -0500 (EST)
- References: <200411070603.BAA18066@smc.vnet.net>
- Reply-to: drbob at bigfoot.com
- Sender: owner-wri-mathgroup at wolfram.com
Highlight any built-in (almost) and press the Help key, and you'll easily find out what it is. I have Needs["Statistics`"] in my Init.m file, so I never load or think about stat packages. I recommend you do the same. I do the same with graphics by including Needs["Graphics`"]. It saves a lot of time.
>> Since this the same as one of your other suggestions, how can using this version of Frequencies be faster?
I hadn't noticed they were the same; I assumed Frequencies was compiled and optimized.
Looking back on the timings again, I notice brt4 and hanlonTreat are within 4% on the largest arrays, with neither of them always winning.
They are the two fastest solvers, and the difference seem to be random noise.
All that being so, I would use the version with Frequencies.
Bobby
On Sun, 7 Nov 2004 01:03:46 -0500 (EST), Bill Rowe <readnewsciv at earthlink.net> wrote:
>
>
> On 11/6/04 at 2:08 AM, drbob at bigfoot.com (DrBob) wrote:
>
>> And the new winner -- for both speed and simplicity -- is:
>
>> brt4[v_List] := Frequencies@Split[v][[All, 1]]
>
> Where is the function Frequencies to be found?
>
> If it is the function found in Statistics`DataManipulation` then that is coded
>
> {Length[#],First[#]}&/@Split[Sort[list]]
>
> Since this the same as one of your other suggestions, how can using this version of Frequencies be faster?
> --
> To reply via email subtract one hundred and four
>
>
>
>
--
DrBob at bigfoot.com
www.eclecticdreams.net
- References:
- Re: Counting Runs
- From: Bill Rowe <readnewsciv@earthlink.net>
- Re: Counting Runs