03-28-2017 10:01 AM
I am trying to create a scatter plot showing correlation stats that includes a best fit line. I am confused as to if I need to do a regression instead?
Also, my data is widely distributed, is there a way to add a log scale (Y axis) to the scatter plot so it is more spread out and can be seen clearer?
03-28-2017 10:07 AM - edited 03-28-2017 10:17 AM
PROC REG and PROC CORR and PROC SGPLOT (and probably others) will create this plot for you. No, you don't need to do a regression, but you can if you want to. The correlation best fit line actually corresponds to the first axis in principal components analysis, you could figure out what that is and then plot that line.
Regarding the log scale, you could easily convert the y-variables to log scale in a data step before doing the plotting.
03-28-2017 10:12 AM
PROC SGSCATTER - SAS Note + sample code for scatter plot + regression line.