Tree Seeds
![]() |
|
Instructions: This demonstration uses a series of scripts that are saved as data table properties. The first one, Count by Weight, simply launches a Bivariate platform on Seed Count and Seed Weight so that you can see a scatter plot of the original data. Note that there is a strong relationship between these two variables but it is not linear. You could use Fit Spline to see the relationship better but instead you will transform the variables to see if you can find a linear relation. The second property, Transformed X and Y, launches a Bivariate platform with the Log Count and Log Weight and adds a Fit Line analysis. You can see that now the linear relationship is quite clear. The third one, Fit Special, launches another Bivariate platform and uses Fit Special with the natural logarithm transformation on the original variables. (Note the difference between log base 10 and natural log changes the intercept estimate.) The Fit Special feature of the Bivariate platform lets you explore other models without transforming the data first. Finally, you can use the last property, Combined, to open a window that shows both approaches side by side.
Requirements: Tree Seeds.jmp (included)


