| Author |
 Topic  |
|
|
sfe10
France
Posts |
 Posted - 07/11/2005 : 10:59:06 AM
|
Hello,
I've some difficulties in using fitting tool on Origin 6.0 like following:
1. I've two colums X and Y, i plot y versus x. x axis and y axis in linear type.
2. I like to find a fitting function of this graph. In this case, i choose polynomial fitting tool by clicking on fit_polynomial button. It shows me a formula (f1) which seems to be fitted well with the data. But when i change the y axis type (from linear to log10, x axis type is always linear), clicking again on fit_polynomia button, it gives me an another formula (f2)with a curve which also seems to be fitted better with the data and when i want to verify this one, this formula (f2)doesn't give me good y data.
Please tell me how can i understand this formula (f2) when i change the type of y axis?
Thanks alot.
Good day |
|
|
easwar
USA
1964 Posts |
Posted - 07/11/2005 : 12:05:31 PM
|
Hi sfe10,
When you do the first fit, the fit forumula corresponds to a polynomial of y as a function of x. When you change the y scale to log10 and do the fit again, the fit formula corresponds to a polynomial of log10(y) versus x.
Also, note that the values of coefficients pasted to graph etc could be limited by numeric precision settings. It is better to use the actual fit coefficients stored in internal variables. When you do a polynomial fit, right after the fit, the coefficients are stored in the stat object in variables stat.pr.a, stat.pr.b1, stat.pr.b2 etc.
You can use these variables stat.pr.a etc to set col values and reproduce the numbers. For more help with this, contact tech support.
Easwar
OriginLab
|
 |
|
|
sfe10
France
Posts |
Posted - 07/11/2005 : 12:52:32 PM
|
Thank easwar very much,
In my understanding of your answer once i do the fit with y axis in log type and x axis in linear type, i'll get a formula displaying on the graph (i've chosen "show formula on graph" option) like this: y=a+b*x+c*x^2....
In reality ( linear axis), this formula must be understood like this:
log10(y)=a+b*x+c*x^2....
so:
y=10^(a+b*x+c*x^2....). and i use this formula to verify the precision of fitting procedure, is it right?
Thanks again! |
|
|
|
easwar
USA
1964 Posts |
Posted - 07/11/2005 : 2:20:58 PM
|
Hi sfe10,
Yes, in this case it would have been less confusing if the formula did say log10(y)=f(x) instead of y=f(x). The formula is for the log10(y) values as you can verify by computing using the coefficients.
Easwar
OriginLab
|
|
|
|
greg
USA
1378 Posts |
|
| |
Topic |
|
|
|
polynomial fitting tool!