Data fitting - Troubleshooting

When the fitting is not following the curves it is necessary to review all the steps from measuring the curves, preparing the curves for fitting, choosing the model to fit to the curves and the result of the fitting. Each step must be optimized such that there is the best change to get reliable results. However, sometimes the fitting will not give a reliable answer due to unknown variables. In these cases it is better to rethink your experimental set-up than reporting improper results.

Experimental set-up

For the best results refer to the Best results section and the Experiments section.

Preparing the curves

Special attention must be paid to the data processing to avoid introducing unwanted artefacts. Please look at the Data processing section

The model

Base the model on the expected interaction but always start with the simplest model possible. Set the variables to global or local depending on the data set. Try to fit with different initial values to make sure that the result of the fitting is reproducible.

Fit inspection

After the fitting is completed, compare the fit with the measured curves. Is the fit following the curves? Where are the deviations? How are the residuals? and not least are the fitted parameters possible with the curves?