I've been working with some growth data and trying to predict some missing data. Ordinarily, polynomial models fit my data best; however, I've found that a few variables are best fit by "distance weighted least squares" in a scatter plot:
Q1: How do I find the equation coefficients for this model to use to predict data?
Q2: Is there a procedure in the software to test the fit and calculate the residuals for this line?
PS, I'm using v10
Are you fitting the data using the 2D scatterplot?
Under 2D scatterplot dialog, you can check the option "Regression (fit) equation" to have the regression equation displayed in the 2D scatterplot.
But the equation displayed is only for linear fitting. I don't think there is an equation available to be displayed for other curve fittings under this 2D scatterplot dialog, like polymomial, exponential or distance weighted least square. Those non-linear fitting in scatterplot is mostly for visualization purpose to check if your data is linearly or non-linearly fitted.
If your data seems to be linearly fitted, you can go to "Statistics | Multiple Regression", it will perform a much detailed linear regression analysis for your variables. You can find regression coefficients, residuals, goodness of fitness, and it can do predictions for you as well if you can provide the values of the independent variables in our regression model.
If you found your data might be bested fitted non-linearly, you can go to "Statistics | Advanced Models | Nonlinear Estimation or General Linear/Nonlinear". There you can find different non-linear estimations/regressions available to fit your data.
Though current version 13 has been improved/changed quite a lot compared to older version like 10, but I believe you can still be able to find those multiple regression and nonlinear estimations/regressions in your version.