[gentoo-user] Re: Modelling software - free - preferably easy to install under Gentoo.

[gentoo-user] Re: Modelling software - free - preferably easy to install under Gentoo.

[James]

Steve [Gentoo] <gentoo_steve <at> shic.co.uk> writes:

> 
> I have some (say 100) discrete data sequences sampling a single analogue
> system with time-stamp data.
> 
It's unclear what you are after. Advice on which mathematical approaches
will work or which software contains those mathematical approaches?

Matlab is the standard for mathematical analysis of all sorts of
phenomenon, from a mathematical perspective.

> I would like to do some analysis on these signals to see if there are
> any interesting things that can be demonstrated - for example, if I
> could show a strong correlation in the signals between two times, but
> none at other times, I might be able to conclude that there was
> communication of some description, but only for a fixed duration.

Very unclear what you are saying. Are these signals related to events in
your network? More information will help.

> 
> At the moment I'm open minded about what kind of software I'd want to
> employ - and also about what I'd like to prove.  Essentially, I'd like
> to analyse the data for features - then ask if they correspond with
> system events I'm already broadly aware about (rather than vice-versa.)

> Can anyone point me in the right direction, please?

You might want to
'cd /usr/portage' and then pick a dir...

'cd sci-mathematics'   and emerge some software who's description
you find potentially interesting....

for example

'exi octave' reveals:

* sci-mathematics/koctave
     Available versions:  0.65-r1
     Homepage:            http://athlone.ath.cx/~matti/kde/koctave/
     Description:         A KDE GUI for Octave numerical computing system

* sci-mathematics/octave
     Available versions:  2.1.57-r1 2.1.69 ~2.1.71-r2 ~2.1.72 2.1.73 ~2.1.73-r1
~2.1.73-r2
     Homepage:            http://www.octave.org/
     Description:         GNU Octave is a high-level language (MatLab
compatible) intended for numerical computations

* sci-mathematics/octave-forge
     Available versions:  ~2004.11.16-r1 2004.11.16-r2 ~2005.06.13
~2005.06.13-r1 ~2006.01.28 2006.03.17 ~2006.03.17-r1
     Homepage:            http://octave.sourceforge.net/
     Description:         A collection of custom scripts, functions and
extensions for GNU Octave



hth,


James




-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Re: Modelling software - free - preferably easy to install under Gentoo.

[Steve [Gentoo]]

On Sat, Jun 16, 2007 at 05:22:47PM +0000, James wrote:
> Matlab is the standard for mathematical analysis of all sorts of
> phenomenon, from a mathematical perspective.

I'm familiar with Matlab... you're the second person to mention
Octave...

> > I would like to do some analysis on these signals to see if there are
> > any interesting things that can be demonstrated - for example, if I
> > could show a strong correlation in the signals between two times, but
> > none at other times, I might be able to conclude that there was
> > communication of some description, but only for a fixed duration.
>
> Very unclear what you are saying. Are these signals related to events in
> your network? More information will help.

I agree - Not only was my post unclear, but I'm unclear about what I
want too. :-)

My data, in  reality, consists national statistics - and my
self-appointed challenge is to establish if, subject to appropriate
analysis, they will expose undocumented trends or other anomalies.  I
don't know what trends or anomalies I want to find until I discover
them... but I suspect that, once found, they'd be interesting. :-)

> 'exi octave' reveals:

Octave is a good suggestion - but probably not what I need.  I've been
pointed at "R" ( http://www.r-project.org/ ) which looks more hopeful,
though I can't find it in portage. If there were an interactive GUI to
apply standard statistical analyses to data as a front-end to R, then
that would likely be just what I want.  Failing that - just finding R
in portage would be a step forwards.

I'd be very interested to know if R has competition...
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Re: Modelling software - free - preferably easy to install under Gentoo.

[Abraham Marín Pérez]

Steve [Gentoo] escribió:
> On Sat, Jun 16, 2007 at 05:22:47PM +0000, James wrote:
>   
>> Matlab is the standard for mathematical analysis of all sorts of
>> phenomenon, from a mathematical perspective.
>>     
>
> I'm familiar with Matlab... you're the second person to mention
> Octave...
>
>   
>>> I would like to do some analysis on these signals to see if there are
>>> any interesting things that can be demonstrated - for example, if I
>>> could show a strong correlation in the signals between two times, but
>>> none at other times, I might be able to conclude that there was
>>> communication of some description, but only for a fixed duration.
>>>       
>> Very unclear what you are saying. Are these signals related to events in
>> your network? More information will help.
>>     
>
> I agree - Not only was my post unclear, but I'm unclear about what I
> want too. :-)
>
> My data, in  reality, consists national statistics - and my
> self-appointed challenge is to establish if, subject to appropriate
> analysis, they will expose undocumented trends or other anomalies.  I
> don't know what trends or anomalies I want to find until I discover
> them... but I suspect that, once found, they'd be interesting. :-)
>
>   
>> 'exi octave' reveals:
>>     
>
> Octave is a good suggestion - but probably not what I need.  I've been
> pointed at "R" ( http://www.r-project.org/ ) which looks more hopeful,
> though I can't find it in portage. If there were an interactive GUI to
> apply standard statistical analyses to data as a front-end to R, then
> that would likely be just what I want.  Failing that - just finding R
> in portage would be a step forwards.
>
> I'd be very interested to know if R has competition...
>   
R is indeed in portage, you may have not found it due to it being 
uppercase. Try the following:

emerge -pv R              # Remember: type 'R' and not 'r'

Also, R has commercial alternatives, such as S and S+, but I'm sure you 
prefer the free version ;-).

HTH,
Abraham

-- 
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Modelling software - free - preferably easy to install under Gentoo.

[Kent Fredric]

On 6/17/07, Steve [Gentoo] <gentoo_steve@shic.co.uk> wrote:
> I have some (say 100) discrete data sequences sampling a single analogue
> system with time-stamp data.
>
> I would like to do some analysis on these signals to see if there are
> any interesting things that can be demonstrated - for example, if I
> could show a strong correlation in the signals between two times, but
> none at other times, I might be able to conclude that there was
> communication of some description, but only for a fixed duration.
>
> At the moment I'm open minded about what kind of software I'd want to
> employ - and also about what I'd like to prove.  Essentially, I'd like
> to analyse the data for features - then ask if they correspond with
> system events I'm already broadly aware about (rather than vice-versa.)
>
> Can anyone point me in the right direction, please?


Not exactly sure what your asking for, but if the data can be
represented as an audio stream of some description you may want to
look at baudline, its a great tool,  but not in portage.

Basicaly an  FFT time/frequency analysis tool

http://www.baudline.com/

If its of no use to  you, It will probably still have the 'oh thats so
cool'  attributes :)
-- 
Kent
ruby -e '[1, 2, 4, 7, 0, 9, 5, 8, 3, 10, 11, 6, 12, 13].each{|x|
print "enNOSPicAMreil kdrtf@gma.com"[(2*x)..(2*x+1)]}'
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Modelling software - free - preferably easy to install under Gentoo.

[Dede]

Hi Steve,

Personally I really like Numpy/Scipy:
http://www.scipy.org/
with Matplotlib for 2D graphs. They have all ebuilds for Gentoo but you
need to edit your /etc/portage/package.keywords to emerge.

I understand your feeling, if you do not really know what you
need, the batteries included of Python are for you. Moreover Python has
R bindings so it's a very flexible solution.

Best regards,

Dede


On Sat, 16 Jun 2007 17:15:24 +0100
"Steve [Gentoo]" <gentoo_steve@shic.co.uk> wrote:

> I have some (say 100) discrete data sequences sampling a single
> analogue system with time-stamp data.
> 
> I would like to do some analysis on these signals to see if there are
> any interesting things that can be demonstrated - for example, if I
> could show a strong correlation in the signals between two times, but
> none at other times, I might be able to conclude that there was
> communication of some description, but only for a fixed duration.
> 
> At the moment I'm open minded about what kind of software I'd want to
> employ - and also about what I'd like to prove.  Essentially, I'd like
> to analyse the data for features - then ask if they correspond with
> system events I'm already broadly aware about (rather than
> vice-versa.)
> 
> Can anyone point me in the right direction, please?
> 
-- 
gentoo-user@gentoo.org mailing list