I am trying to extrapolate data from Blackwell Contrast Threshold 1946, however when i use TableCurve3D, the result is not consistent and unpredictable.
Is there something that I am missing?
I use the angular subtense as X, Bo (Surface Brightness) as Y, and Contrast Threshold as Z.
I have attached the blackwell data and his full paper.
Please help me guys.
You can use Excel Solver.
You need data to fit, an equation with parameters to fit to. Make some initial guesses to the parameters and use these parameters to calculate data values you predict. Get the error by subtracting each of the actual values from the predicted values and squaring. Add all the squared errors to get the sum squared error. The sum squared error is what you minimise using solver.
It is easier to interpolate than to extrapolate outside the bounds of your actual data.
see http://people.chem.ucsb.edu/laverman/leroy/Chem116/PDF116CL/Solver.pdf
and for multiple independent variables see
http://cob.jmu.edu/wangpx/kj/ms/2reg/lectureregression.pdf
If you have a data distribution and looking for the best fit distribution or the variables of the fit curve, I recommend two software which are Matlab and EasyFit.
- Badges
- Science topic
- Similar topics
- Thesis Research