Friday, March 21, 2008

My R-Square is bigger than your R-Square

Just remember, when you're forecasting demand from historical data that includes trend, level and seasonality -- you darn well better be sure that your independent variables cut the mustard or someone like me is going to have a different forecast result and as we all know, he who has the bigger R-Square wins. And, FYI, if you have multiple sets of demand and need to find out which of them, if any, contain trend, level and/or seasonality, I would recommend using a bivariate correlation to measure significance on all variables in one shot rather than doing things the long way and running multiple linear regressions until one of these pops up. (Yes Mike, that means you.)

That pretty much sums up my life for the day. No kidding. That's what I did. And my R-Square was bigger than Kyle's so we ended up using my forecast results for the assignment instead of his.

Perhaps my father is the only who understands what this all means which makes the picture below all the more amusing.

My mom took this picture while my parents were here last week. It begs for a humorous caption -- specifically one that reflects what Phoebe is thinking. Please post whatever caption you think fits. The ones we have so far are:

"Who ARE these people? And when will Melba come home and take me on a walk?" -- Vernon P.

"8 Billion people in the world -- and I get stuck with these two?" -- Melba

"How do I get online?" -- Grandma Dewey

"Hey, a quarter!" -- My brother Jeff

"Well, it could be worse. At least these guys aren't lawyers." -- Vernon P. (that one is for you, Jeremy.)

"Phoebe's the only one in the picture I can relate to. It's the story of my life." Larry (Mom Boyer)

And this is my favorite -- from my sister:


jkidd said...

I personally liked Melba's "8 billion people" caption best. Oh, and Sammie, I understood exactly what you were saying, and you should always remember that you can always increase your R-squared simply by adding more variables, so that is why most statisticians and economists use R-bar-squared, which adjusts for the number of variables you have included, and punishes you for using extraneous and irrelevant variables. See, I'm a lawyer and a geek.

Sara Corinne said...

Yay I win.