Board 8 > Does anyone here know how to do linear regression?

(+) Yeah! (+)

I'm really at a bit of a loss here and I'm in over my head. If anyone has some time and can offer me some guidance, that'd be appreciated greatly.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

possibly

--
senorhousemouse

****. This reminds me I have to look up how to do this crap again since I'm probably TAing a course on numerical methods.

--
assert(!hotterThan(foo, "Hot Nymphomaniacal Lesbian Mind-Controlling Dominatrix Fairy Doctors with glasses"))

isn't linear regression basically just "line of best fit"? It's been 3 years since I took stats, but I'm pretty sure that's it. Stuff with r and r squared values.

--
~AHylian is Still Alive~

do you have a graphing calculator or excel?

--
~AHylian is Still Alive~

Oops. I meant multiple regression. I actually got the linear regression down

I have excel, and a file full of numbers that I'm having a hard time understanding. I'm piecing it together but I have a 9pm EST deadline and I don't know if I will put it together fast enough.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

Here's the data I've found relevant so far



Here are my q's to be answered

State the multiple regression equation.
Interpret the meaning of the slopes in the equation.
Predict the salary for a 40-year-old male employee with 5 years of experience.
At the .05 level of significance, determine whether each explanatory variable makes a significant contribution to the regression model.
Interpret the meaning of the r-squared.
Interpret your findings. What changes would you make to the model, given the results?

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

I've been working out how to get the slopes, but I'm still pretty bad at this. This is actually my first math class in a decade.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

Oh man. I wish I could help more, but I haven't taken stats in so long.

R squared is, from what I remember, the percent of y explained by x.

I wish I could help more, but 5 minutes isn't enough time to jog my memory :/

--
~AHylian is Still Alive~

s(b) is the slope, no?

--
~AHylian is Still Alive~

I think that is standard deviation and b is slope, actually.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

Hmm. I swear b is the y=mx+b, but I could be wrong.

--
~AHylian is Still Alive~

I only kind of glanced at the picture, didn't fully understand all of it.

--
~AHylian is Still Alive~

But I'm reeeeaaally confused at this point, so I dunno.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

Found a sample in the textbook and it looks like b = coefficients and s(b) = standard error.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

Can't you use a pseduo-inverse technique with matrices to get the line of best fit?

--
_foolmo_
'Most people at least try to say something funny. See foolmo's post as an example.' - The Real Truth

So what I'm seeing is b's intercept is what the result would be if everything else were zero. If we hold experience and gender constant then each increase of 1 in age adds 2098.xx to the b's intercept... things are clicking just a tad...

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

foolmor0n posted...
Can't you use a pseduo-inverse technique with matrices to get the line of best fit?

I have absolutely no idea how to do that, and probably shouldn't.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

and then y = b0 + b1(x)

--
ND

b1 = ((xi - xbar) * (yi - ybar)) / (xi-xbar)^2
b0 = ybar - (b1 * xbar)

where xi is each instance of x, and xbar is the average. same with y

--
ND

oh whoops, three variables. should have checked the spreadsheet >_>

--
ND

Yeah, that was my fault for the incorrect title.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

do you have minitab? if not, download it.

very helpful for multiple regression.

--
ND

Going to do that right now, then. Thanks.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!

And there are problems with making an account. Oh well.

--
"Principally I hate and detest the animal called man, although I heartily love John, Peter, Thomas and so forth" - Jonathan Swift
BT with the victory!