/anscombe_and_influence

Lecture on the fun of residuals

Primary LanguageSAS

anscombe_and_influence

Lecture on the fun of residuals Anscombe's Quarted is four datasets that each yield the same exact regression results, however three of the data sets are seriously flawed. The lesson is that just because a regression looks good, e.g., has a high R-squared, does not mean its not worthless.

I think the Anscombe data and its lessons are fun, hence i refer to this as the fun dataset.

All use should cite Dr. Steven C. Myers, econdatascience.com

Extended citation: Associate Professor Emeritus, The University of Akron.

FILES

Results_Influence.sas This file is the sas code that will explore the data and the use of influence statistics to discover data anomomlies after the regression is run.

Results_Influence.pdf This file includes the output from the sas code in a pdf format

fun_results.xlsx This file includes a table of results prepared for a formal paper (Journal style) and much more.