| T O P I C R E V I E W |
| physiologist |
Posted - 08/02/2012 : 05:24:57 AM
Origin Ver. and Service Release (Select Help-->About Origin): 8.6G 32 bit
Operating System: Win7 Prof
Hello,
I have a problem fitting my data with a third order exponential fit.
I have 1000 datapoints over a time of 20 seconds. I smoothed the datapoints and performed a fit. The fit of my curve looks good, but the decay constants dont appear to be right.
For example, t1 has a constant of 1,43*10^7 and a standard error of 8,6*10^8. Also, t3 has a smaller than t2...
How can I get values that resemble the values in the graph?
Is the fit over 1000 datapoints too much for origin to handle correct contants?
Would be nice if someone could help me.
thanks in advance |
| 3 L A T E S T R E P L I E S (Newest First) |
| Sam Fang |
Posted - 08/06/2012 : 01:41:46 AM
For the ExpDecay3 fitting, it seems x0 in the fitted result is not correct, maybe you can set a bound for it or fix it. If you can estimate the range for the decay constants, you can try setting bounds for them.
You can also send us your fitting data. We can try it.
To upload a file, click Send File to Tech support button in the top right of the forum.
Thanks.
Sam
OriginLab Technical Services |
| physiologist |
Posted - 08/02/2012 : 11:24:14 AM
i just checked for the comparison tool but its only available in the pro version.
also, at a 2 exp fit the transition from the high to a smaller decay isnt described.
to determine wheter a 2exp or a 3exp fit is better, i analysed the residuals.
so when a 2exp doesnt describe my data and 3exp is over-parameterized...what is the best way to describe my data?
also for a 2exp fit the parameter values also seem off.
 |
| easwar |
Posted - 08/02/2012 : 09:42:12 AM
Hi,
The number of data points is not the problem.
For your data, the model you chose may be over-parameterized...there are more parameters than necessary, so the iteartive fitting procedure cannot find a unique set of parameter values, thus leading to large error estimates on some parameters.
Also, you x data has no offset, starts at zero, so try the expdec2 function instead of expdecay3.
You can fit the data with two different functions, and then use the Analysis->Fitting->Compare Models tool to compare the two fits, to get a statistical conclusion as to which model is more likely to represent your data.
Easwar
OriginLab |
|
|
