Paul we can agree that there are instances in product development where focusing in a narrow subset of the data (say 27%) can lead to tangible improvements.

But you're being disingenuous to say that's the case here.

In your story we have indicators (urban centers) that link the 27% of "bad" data to a defined problem (inaccurate GPS placement).

Let's just say winning the MAC Championship in 2015 is the defined problem. In the 2015 pre-season you look over the past three seasons to see if OHIO is likely to solve that problem in 2015.

But when you exclude 73% of the data how can you differentiate a team like Eastern Michigan who did not beat conference opponents with winning records over that period and did not beat out-of-conference opponents with winning records?

On paper, EMU and Ohio are the same team when you look at 27% of the data. As a consultant, would you really have the same recommendation for EMU and Ohio? (No)

What would be your basis for a different recommendation for EMU and Ohio? (the other 73% of data)

The way to estimate if Ohio has a better chance than EMU (or any other MAC team) to win the MACC in 2015 is to look at their probability of Ohio winning enough conference games to get to the championship game.

Although Ohio did not beat MAC teams with winning records over that period they did beat out of conference teams with winning records that would indicate it certainly was possible Ohio could beat at least some MAC teams with winning records.

And let's be clear, "good" MAC teams are worth the same win as "bad" MAC teams. Out of the East Ohio needed 4 of them in 2008 (Ohio got 3). Ohio needed 7 out of the East in 2010 (Ohio got 6). Last year Ohio needed 5 (Ohio got 4). And once you get to MAC Championship game it's a crap shoot, Ohio can't control their opponent. CMU was a buzzsaw. NIU has won and lost, Ohio has blown a 20 point 2nd half lead, heck then undefeated Ball State lost to Buffalo.

Every year the win requirement to get to the MAC Championship changes. Ohio's got 2 so far. Some here may say Ohio won't get enough conference wins in 2015. And what I'm saying is you need more than 27% of the data to state that hypothesis with intellectual integrity.

You want to be old after 42 years, Keep dropping the hammer and grinding the gears

Judging by the time-stamp, I'm inferring this is somehow in connection to your feeling of vindication?

I'll leave one last reminder. The original post was in no way, shape, or form written to say would should be done. The original post was written to show how not to do it (self-serving data manipulation).

ARE YOU LITERALLY INSANE?!

I'VE STATED EXPLICITLY IN THIS THREAD, MULTIPLE TIMES, THAT I'M OPEN TO EVERY SINGLE GOD #### POINT OF VIEW ON WHAT ACTIONS COULD BE TAKEN. EVEN YOUR PATHETIC EXCUSE FOR COGNITIVE REASONING.

THE ONLY INCREDIBLY SMALL, NARROW, SIMPLE POINT I HAVE BEEN TRYING TO MAKE IS THAT YOU CAN'T EXCLUDE 73% OF THE DATA FROM THE DATA SET YOU USE TO PROVE YOUR POINT.

YOU CAN'T DO THAT.

YOU CAN'T DO THAT IN SCHOOL AND YOU CAN'T DO THAT WHEN SOMEONE'S PAYING FOR YOUR ANALYSIS.

YOU'RE LITERALLY INSANE.

Okay.

In that case, what do you think of how President Gore did in his second term as President?

"And scene! Reminder that Mr. Sorkin would like everybody on set bright and early tomorrow at 6 a.m."