Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronic Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

[Needed]Curve Fitting for a a set of datas, links

Status
Not open for further replies.

vamanan

Full Member level 2
Joined
Aug 9, 2005
Messages
135
Helped
10
Reputation
20
Reaction score
4
Trophy points
1,298
Location
India the Great
Activity points
2,286
[Needed]Curve Fitting

Hi,
Can anybody explain me how curve fitting is done for a a set of datas.
You can guide me by giving links also.
Regards,
Vamanan
 

evision

Newbie level 3
Joined
Jan 7, 2009
Messages
3
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,297
Re: [Needed]Curve Fitting

Take an inverse of (total steps -1), raise it to the ratio of Last value/first value
 

Kral

Advanced Member level 4
Joined
Mar 28, 2005
Messages
1,326
Helped
280
Reputation
558
Reaction score
85
Trophy points
1,328
Location
USA
Activity points
13,418
Re: [Needed]Curve Fitting

vamanan,
Two common methods of curve fitting are polynomial curve fitting, and least squares curve fitting. In polynomial curve fitting, a polynomial is found for which each of the data points occurs on the polynomial curve. In least squares fitting, a polynomial is found for which the sum of the squares of the errors between the data points and the polynomial is minimal. See the folowing:
. http://en.wikipedia.org/wiki/Curve_fitting
. http://en.wikipedia.org/wiki/Linear_least_squares
Also, Google on l"east squares" and "polynomial curve fit". You'll get lots of hits.
Regards,
Kral
 

iyami

Full Member level 4
Joined
Apr 12, 2008
Messages
210
Helped
60
Reputation
120
Reaction score
40
Trophy points
1,308
Location
Earth (or at least somewhere nearby)
Activity points
2,728
Re: [Needed]Curve Fitting

In least squares fitting, a polynomial is found for which the sum of the squares of the errors between the data points and the polynomial is minimal.

Just a small addition. These functions need not be polynomials. You may also have data which
you know is generated by a sum of sine functions or exponentials. In that case you can use
(non-linear) least-square fitting to find the best (or at least a close enough) fit among functions
of this type.

The general procedure consists of the following steps
1) decide which type of functions you want to use as candidates
2) decide how you define the distance between the data and these functions
3) find a close-enough functions from these candidates.

1) are often polynomials of a fixed degree, ration functions, sine waves etc.
2) usually l2-norm (Euclidean distance for functions)
3) there are many open source libraries/programs for linear/non-linear least square fitting
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Top