regression tools?

[Justin7]

I have 2000 data points, with 5 variables each.

I want to curve fit the first 4 to compute the 5th. Does anyone have software that can do this easily?

[Ganchrow]

I have 2000 data points, with 5 variables each.

I want to curve fit the first 4 to compute the 5th. Does anyone have software that can do this easily?

Excel's LINEST() and LOGEST() functions performs linear and logarithmic regressions, respectively.

[Justin7]

Ganch,

Maybe you can speed me through this.

I've imported my data into an excel sheet. A2-D2 are inputs, which return E2.
This goes down to about A20000 to D20000, with the return of E20000.

I'd like to find the equation E = Const + Am1 + Bm2 + Cm3 +Dm4, with the smallest error. I'd prefer linear error, but mean squared works too. What is the syntax for that?

[MrX]

LINEST(E2:E2000,A2:D2000,TRUE,TRUE)

LINEST is an array function, so you need to highlight an area 5 columns by 5 rows with the LINEST formula in the upper left. Then hit F2 and then CTRL-SHIFT-ENTER.

See the help entry on linest for an explanation of the output.

[Justin7]

Um, I'm dumb. Can you show me on this data file?

[MrX]

Here ya go.

[Justin7]

Thanks. Now i just have to figure out why predicted scoring goes down as your passing efficiency goes up

[MrX]

Thanks. Now i just have to figure out why predicted scoring goes down as your passing efficiency goes up

Yikes.

Are you sure you're interpreting the LINEST columns correctly, 'cause their order is kinda counter-intuitive? The far left coefficient corresponds to the rightmost data column (column D in your example).

[Justin7]

That would explain it. I need to swap my A columns with D, and B with C.

Is there a way to do this with the linest command?

Btw, I really appreciate your help on this. This is a monster project I'm working on, but it will be worth it when NCAAF starts.

[Justin7]

Actually, I don't need to resort it. I can just reverse the order that I read the coefficients.

[MrX]

Actually, I don't need to resort it. I can just reverse the order that I read the coefficients.

There ya go!

Left to right the coefficients correspond to Column D, C, B, A, Constant.

[Justin7]

Btw, the results were spectacular. The first pieces are slowly fitting together for the NCAA model.

[bookie]

Justin did you find a good datascraper?

[Justin7]

Justin did you find a good datascraper?

I paid 3k for 10 years of comprehensive NCAA data awhile ago. I have a scraping agreement for this fall - data in exchange for my plays before post them.

[durito]

I paid 3k for 10 years of comprehensive NCAA data awhile ago. I have a scraping agreement for this fall - data in exchange for my plays before post them.

How comprehensive?

Is this something I could buy somewhere?

[Justin7]

How comprehensive?

Is this something I could buy somewhere?

Most of the things in a box score.

You could hire a programmer, trade, or buy it off anyone that will sell it. Lots of pros have databases like this.