I will admit. I never met a scatter plot I didn't like. Unfortunately all scatter plots are not created equal. My point is--we need to have conversations about data in the language we can all understand. The graph shows unequal variability of a variable across the range of values that predict the values--heteroscedasticity. Basically, perhaps these represent subpopulations that have a different rate of variability. This matters because regression assumes constant variance across all observations. |
Are you a data professional? If you aren't speaking in language we can understand--It isn't me it is you.