Hi there,
I am using a multiple power regression model based on the following equation:
y=a*(x1^b1)*(x2^b1)
I have determined the "a", "b1" and "b2" parameters using Excel after applying a logarithmic transformation in the variables (dependent and independent)
ln(y)= ln(a) + (b1*lnx1)+(b2*lnx2)
The results are shown in the Figure. However, they refer to the logarithmic model. Now, I want to associate them to the power model. Can I do the exponential transformation of all values?
Also, to determine the quality of the fitting, can I use the R Square value? Even though it is a nonlinear regression model. Or use the standard error of estimation?
If using the standard error, would it be the exponential of the value calculated by Excel (highlighted in yellow)?