| T O P I C R E V I E W |
| ghostyblop |
Posted - 11/23/2024 : 5:01:00 PM
Hello,
I've been fitting a 2D gaussian to some of my data. My reported reduced chi-squared fit is 0.000144 which is very small.
I have read on a previous forum https://my.originlab.com/forum/topic.asp?TOPIC_ID=4058
that the reported chi-squared fit will be small in a perfect fit since chi^2 = 0 (and then this is subsequently divided by DOF). However, online I am reading if a chi-sqaured fit is < 1 than this indicates an overfit.
As a result, I am confused as to whether a very small value below 1 is an indication of a good fit.
Any help on this topic would be great, Thanks
|
| 1 L A T E S T R E P L I E S (Newest First) |
| YimingChen |
Posted - 11/25/2024 : 3:43:32 PM
"a very small value below 1 is an indication of a good fit."
This is true only if the measurement error is well-defined and incorporated as weights in the fitting. For more details, please refer to the discussion section of the Wiki page and the Origin webpage on how weights are included in fitting.
https://en.wikipedia.org/wiki/Reduced_chi-squared_statistic
https://www.originlab.com/doc/origin-help/nlfit-theory#Weighted_Fitting
Thank you
James |
|
|
