| T O P I C R E V I E W |
| physiologist |
Posted - 09/02/2012 : 12:29:09 PM
Origin 8.6 Pro
Win7
Hello Originforum,
i would like to ask a little question on a fitting problem I have. I have 1000 datapoints over a timeperiod of 20 seconds. I smoothed the data and the result is the attached graph. This graph can be described by the "3expdec" function from time point 0 to aprox. 8000ms.
But the function also describes the graph for example from 0 to 7000ms. The rest of the graph can be described by the "1expdec" function. So my question is, how do I determine appropiate time ranges??? Is there any mathematical way to do this? Setting this "by eye" seems way to inaccurate.
Thanks for the help in advance
|
| 2 L A T E S T R E P L I E S (Newest First) |
| greg |
Posted - 09/04/2012 : 09:12:36 AM
This could be handled by using script - see http://wiki.originlab.com/~originla/ltwiki/index.php?title=LabTalk:Curve_Fitting for some examples. The idea would be to select test fitting ranges, do the fits, then move the ranges and try again. But coming up with the code that does that is only the beginning.
Your next problem is determining where the best fits occur. While you can compare expdec1 to expdec1 fits (and expdec3 to expdec3 fits) for two ranges, you cannot compare an expdec1 fit to an expdec3 fit.
Add to this that there must be some transition zone over which neither function exactly fits unless you know of some real event that occurs at some specific time to change the dynamics.
In the end, by eye may be just as good - and certainly quicker - than attempting the above which would include a number of dubious qualifications. |
| physiologist |
Posted - 09/02/2012 : 12:33:04 PM
editing and image upload dont seem to work
here is the image I wanted to add.
http://s1.directupload.net/file/d/3001/3226u8qt_jpg.htm |
