Minitab

So when I'm done, I have 59 points.

What's the definition of what's the best fit line in a regression? When you run a regression on Minitab or Excel, what does it do? It's sum of least squares, but tell me in terms.

Basically, what you do is you try to fit a line through where the square distances from the line are minimized.

If you ever get a chance, go back and look at how regressions were done in the 60s and the 70s.

You had to square the distance and minimize it.

Thank God we don't live in the 60s and the 70s.

But that's what that Excel package or Minitab is doing.

It's putting the best fit line.

[laughs]

But yeah, when I started, I was trained on Minitab because Penn State was, like, a Minitab school, and then I got to work-

Oh, old school

... and there was this thing called Excel, and I was like, "I guess this is my life for the rest of the..." But anyway, I just started messing around with it, and I, uh, was very fortunate to work alongside some folks who were getting into it as well.

Most statistics packages.

Is it Minitab that you guys still have here at, at Stern? I don't know what they're l- No, but I, I know they also have SPSS.

I think they have a bunch of different statistics packages you can use as Stern MBA.

They all will give you predictions.

It's not meaningful.

So if you have a traditional statistics package like SPSS or Minitab, there's actually in most of these packages an option of removing the intercept.

It's, as I said, not great practice because it forces the line through, your regression line through the intercept.

But if you have a very negative intercept, you're worried about your predicted values turning negative, all you can do is actually take out the intercept, rerun the regression without a constant, without the intercept, It's not the best fit line anymore because you force the line through, but you now have the problem gone away.