Good luck,&nbs= p;

Cina

-----------------------= -

Cina Forou= tan-Nejad, PhD

CEITEC-Central European Institute of Technology, Masaryk University,&nbs= p;

Brno, Cze= ch Republic

On Fri, 12 Feb, 2016 at 16:28, Andr=C3=A9 Farias= de Moura moura[*]ufscar.br
<owner-chemistry++ccl.net> wrote= :

not really a computatio= nal chemistry issue, but some general remarks could be done here.

<= /div>
I'm not really aware of any program that would guess a fitting fu= nction for you, neither Maple nor any other commercial/academic/free softwa= re.

There's no simple answer for your question, guessing= the function is up to you.

If there are ana= lytical theories for your problem, begin with the theory-derived equations.=

If not, begin with simple functions and inc= rease the number of adjustable parameters judiciously, since the quality of= the fitting improves with the number of adjustable parameters, but physica= l meaning usually becomes blurry after a few extra empirical parameters are= added. And always do the statistical validation of the newly added paramet= ers, since they soon begin to loose statistical significance (if you don't = know how to do that, then you need to take a step back and do some backgrou= nd reading on basic statistics and regression techniques).
=
I hope it helps.

Andre
<= div class=3D"gmail_extra">
On Thu, Feb 11, 20= 16 at 7:28 PM, Safiya Ess Amer amersaf85-,-yahoo.com <owner-chemistry*|*ccl.net> wrote= :

Sent to CCL by: "Safiya Ess Amer" [amersaf85*_*yahoo.com]
Hi all Subscribes
I have data for a system and need to make fitting for these data. Is there = any
program can help me to guess a general form for my data? I have Maple12
program to do fitting by using Least square method, but need initial guess = to
do that.

Best regards

Safiya Amer
amersaf85%a%yahoo.com
Misrata University-Libya

-=3D This is automatically added to each message by the mailing script =3D-=

E-mail to subscribers: CHEMISTRY*|*ccl.net= or use:
http://www.ccl.net/cgi-bin/ccl/sen= d_ccl_message

E-mail to administrators: CHEMISTRY-RE= QUEST*|*ccl.net or use
http://www.ccl.net/cgi-bin/ccl/sen= d_ccl_message

Subscribe/Unsubscribe:
http://www.ccl.net/chemistry/sub_unsu= b.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemistry= /announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/search= ccl/index.shtml

http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instruc= tions/

--
_____________

Prof. Dr. Andr=C3=A9 Farias de Moura
Department of Chemistry
Federal University of S=C3=A3o Carlos
S=C3=A3o Carlos - Brazil
phone: +55-16-3351-8090

------=_Part_5474595_244552397.1455292642092-- From owner-chemistry@ccl.net Fri Feb 12 13:14:01 2016 From: "Ronald Cook cookrl(!)tda.com" To: CCL Subject: CCL: Fitting of Data Message-Id: <-52022-160212122831-29601-7/2Bp5XoFy4in1SXN2ckIg]![server.ccl.net> X-Original-From: Ronald Cook Content-Type: multipart/alternative; boundary=089e0115ecf0963aff052b95fe61 Date: Fri, 12 Feb 2016 10:28:24 -0700 MIME-Version: 1.0 Sent to CCL by: Ronald Cook [cookrl{:}tda.com] --089e0115ecf0963aff052b95fe61 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable I use a combination of Eureqa which is a genetic algorithm type of fitting program which provides a series of fitted equations with increasingly better fits and complexity. Eureqa tends to overfit data (especially for smaller datasets where it is difficult to remove a portion of the data for out of sample testing) I then follow up by subjecting the models developed by Eureqa to Correlated Components Regression (CCR). CCR builds reliable models via a fast algorithm that incorporates M-fold cross-validation. As explained by Dr. Jay Magidson who developed the code at Statistical Innovations, even with high collinearity within the variables, good robust models with high M-fold cross-validation scores are generally found using CCR Ron Cook On Fri, Feb 12, 2016 at 5:58 AM, Andr=C3=A9 Farias de Moura moura[*]ufscar.= br < owner-chemistry|-|ccl.net> wrote: > not really a computational chemistry issue, but some general remarks coul= d > be done here. > > I'm not really aware of any program that would guess a fitting function > for you, neither Maple nor any other commercial/academic/free software. > > There's no simple answer for your question, guessing the function is up t= o > you. > > If there are analytical theories for your problem, begin with the > theory-derived equations. > > If not, begin with simple functions and increase the number of adjustable > parameters judiciously, since the quality of the fitting improves with th= e > number of adjustable parameters, but physical meaning usually becomes > blurry after a few extra empirical parameters are added. And always do th= e > statistical validation of the newly added parameters, since they soon beg= in > to loose statistical significance (if you don't know how to do that, then > you need to take a step back and do some background reading on basic > statistics and regression techniques). > > I hope it helps. > > Andre > > On Thu, Feb 11, 2016 at 7:28 PM, Safiya Ess Amer amersaf85-,-yahoo.com < > owner-chemistry*|*ccl.net> wrote: > >> >> Sent to CCL by: "Safiya Ess Amer" [amersaf85*_*yahoo.com] >> Hi all Subscribes >> I have data for a system and need to make fitting for these data. Is >> there any >> program can help me to guess a general form for my data? I have Maple12 >> program to do fitting by using Least square method, but need initial >> guess to >> do that. >> >> Best regards >> >> Safiya Amer >> amersaf85%a%yahoo.com >> Misrata University-Libya >> >> >> >> -=3D This is automatically added to each message by the mailing script = =3D- >> E-mail to subscribers: CHEMISTRY*|*ccl.net or use:>> >> E-mail to administrators: CHEMISTRY-REQUEST*|*ccl.net or use>> >> >> > > > -- > _____________ > > Prof. Dr. Andr=C3=A9 Farias de Moura > Department of Chemistry > Federal University of S=C3=A3o Carlos > S=C3=A3o Carlos - Brazil > phone: +55-16-3351-8090 > --089e0115ecf0963aff052b95fe61 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable

I use a combination of Eureqa= which is a genetic algorithm type of fitting program which provides a seri= es of fitted equations with increasingly better fits and complexity. =C2=A0= Eureqa tends to overfit data (espe= cially for smaller datasets where it is difficult to remove a portion of th= e data for out of sample testing) =C2=A0I then=C2=A0follow up by subjecting= the models developed by Eureqa to Correlated Components=C2=A0Regression (C= CR). CCR builds reliable models via a fast algorithm that incorporates M-fo= ld cross-validation.=C2=A0 As explained by=C2=A0Dr. Jay Magidson who develo= ped the code at Statistical Innovations, even with high collinearity within= the variables, good robust models with high M-fold cross-validation scores= are generally found using CCR

Ron Cook

On Fri, Feb 12, 2016 at 5:5= 8 AM, Andr=C3=A9 Farias de Moura moura[*]ufsca= r.br <owner-chemistry|-|ccl.net> wrote:

not really a computational chemistry= issue, but some general remarks could be done here.

I&#= 39;m not really aware of any program that would guess a fitting function fo= r you, neither Maple nor any other commercial/academic/free software.

<= br>

There's no simple answer for your question, guessing the = function is up to you.=C2=A0

If there are analytic= al theories for your problem, begin with the theory-derived equations.=C2= =A0

If not, begin with simple functions and increa= se the number of adjustable parameters judiciously, since the quality of th= e fitting improves with the number of adjustable parameters, but physical m= eaning usually becomes blurry after a few extra empirical parameters are ad= ded. And always do the statistical validation of the newly added parameters= , since they soon begin to loose statistical significance (if you don't= know how to do that, then you need to take a step back and do some backgro= und reading on basic statistics and regression techniques).

I hope it helps.

Andre

On Thu, Feb 11, 2= 016 at 7:28 PM, Safiya Ess Amer amersaf85-,-yahoo.com <owner-chemistry*|*ccl.net>= wrote:

Sent to CCL by: "Safiya Ess Amer" [amersaf85*_*yahoo.com]
Hi all Subscribes
I have data for a system and need to make fitting for these data. Is there = any
program can help me to guess a general form for my data? I have Maple12
program to do fitting by using Least square method, but need initial guess = to
do that.

Best regards

Safiya Amer
amersaf85%a%yahoo.com
Misrata University-Libya

-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY*|*ccl.net or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST*|*ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

Subscribe/Unsubscribe:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/sub_un= sub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemist= ry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/sear= chccl/index.shtml
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/ins= tructions/

_____________

Prof. Dr. Andr=C3=A9 Farias de Moura
Department of Chemistry
Federal University of S=C3=A3o Carlos
S=C3=A3o Carlos - Brazil
phone: +55-16-3351-8090

--089e0115ecf0963aff052b95fe61-- From owner-chemistry@ccl.net Fri Feb 12 13:52:01 2016 From: "Brian Skinn bskinn###alum.mit.edu" To: CCL Subject: CCL: Fitting of Data Message-Id: <-52023-160212134443-17899-n5dikaJA3MhSysYXvPD2IQ]^[server.ccl.net> X-Original-From: Brian Skinn Content-Type: multipart/alternative; boundary=001a11415bc01f82ed052b970fba Date: Fri, 12 Feb 2016 13:44:17 -0500 MIME-Version: 1.0 Sent to CCL by: Brian Skinn [bskinn(_)alum.mit.edu] --001a11415bc01f82ed052b970fba Content-Type: text/plain; charset=UTF-8 I have found the Akaike Information Criterion and its corrected variant (used when n < ~40) to be helpful in guiding model selection: https://en.wikipedia.org/wiki/Akaike_information_criterion A more straightforward definition useful for least-squares regression with normally-distributed errors can be found on page 4 of the PDF here: http://avesbiodiv.mncn.csic.es/estadistica/senseaic.pdf For a fixed dataset, if adding a parameter to the fitting model does not decrease the AIC (or AICc, as appropriate), then the new parameter can be considered to provide no additional information to the fit. -Brian On Fri, Feb 12, 2016 at 11:50 AM, Jim Kress jimkress35:+:gmail.com < owner-chemistry_+_ccl.net> wrote: > > Sent to CCL by: "Jim Kress" [jimkress35[*]gmail.com] > R and other programming languages are useful but can involve your time and > effort for coding. Not EVERYONE uses R, BTW. > > Unless you have an underlying theoretical description of your works, you > will have to try to find a function that dose the best job of fitting your > data. > > For fitting data I use: > > Datafit > > http://www.oakdaleengr.com/ > > It's very flexible, does non-linear as well as linear fits and allows you > to > explore a variety of functional forms to fit your data against. > > There are other packages available. > > Jim > > -----Original Message----- > > From: owner-chemistry+jimkress35==gmail.com:-:ccl.net > [mailto:owner-chemistry+jimkress35==gmail.com:-:ccl.net] On Behalf Of > Safiya > Ess Amer amersaf85-,-yahoo.com > Sent: Thursday, February 11, 2016 4:28 PM > To: Kress, Jim > Subject: CCL: Fitting of Data > > > Sent to CCL by: "Safiya Ess Amer" [amersaf85*_*yahoo.com] Hi all > Subscribes > I have data for a system and need to make fitting for these data. Is there > any program can help me to guess a general form for my data? I have Maple12 > program to do fitting by using Least square method, but need initial guess > to do that. > > Best regards > > Safiya Amer > amersaf85%a%yahoo.com > Misrata University-Libyahttp:// > www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt> > > --001a11415bc01f82ed052b970fba Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable

I have found the Akaike Information Criterion and its corr= ected variant (used when n < ~40) to be helpful in guiding model selecti= on:

https://en.wikipedia.org/wiki/Akaike_i= nformation_criterion

A more= straightforward definition useful for least-squares regression with normal= ly-distributed errors can be found on page 4 of the PDF here:

http://avesbiodiv.mncn.csic.es/estadistica/senseaic.pdf
=

For a fixed dataset, if adding a p= arameter to the fitting model does not decrease the AIC (or AICc, as approp= riate), then the new parameter can be considered to provide no additional i= nformation to the fit.

-Brian

On Fri, Feb 12, 2016 at 11:50 AM= , Jim Kress jimkress35:+:gma= il.com <owner-chemistry_+_ccl.net> wrote:

Sent to CCL by: "Jim Kress" [jimkress35[*]gmail.com]
R and other programming languages are useful but can involve your time and<= br> effort for coding.=C2=A0 Not EVERYONE uses R, BTW.

Unless you have an underlying theoretical description of your works, you will have to try to find a function that dose the best job of fitting your<= br> data.

For fitting data I use:

Datafit

http://www.oakdaleengr.com/

It's very flexible, does non-linear as well as linear fits and allows y= ou to
explore a variety of functional forms to fit your data against.

There are other packages available.

Jim

-----Original Message-----
> From: owner-chemistry+jimkress35=3D=3Dgmail.com:-:ccl.net
[mailto:o= wner-chemistry+jimkress35=3D=3Dgmail.com:-:ccl.net] On Behalf Of Safiya
Ess Amer amersaf85-,-yahoo.com
Sent: Thursday, February 11, 2016 4:28 PM
To: Kress, Jim=C2=A0 <jimkress35:-:gmail.com>
Subject: CCL: Fitting of Data

Sent to CCL by: "Safiya Ess Amer" [amersaf85*_*yahoo.com] Hi all Subsc= ribes
I have data for a system and need to make fitting for these data. Is there<= br> any program can help me to guess a general form for my data? I have Maple12=
program to do fitting by using Least square method, but need initial guess<= br> to do that.

Best regards

Safiya Amer
amersaf85%a%yahoo.com
Misrata University-Libyahttp://www.ccl.net/cgi= -bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp:/= /www.ccl.net/spammers.txt

-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY_+_ccl.net or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST_+_ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

Subscribe/Unsubscribe:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/sub_un= sub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemist= ry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/sear= chccl/index.shtml
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/ins= tructions/