PDA

View Full Version : Regression Equation to find Y Intercept?



Reboika
11-11-2015, 02:41 PM
I am a graduate who is looking over past exam papers for statistics. Basically starting from scratch in my learning since I absorbed very little from when I did it in uni (was compulsory for psychology students to take statistics for two of the four years of the degree).
I have a question based on Regression, and I'm struggling to get it. See below. I have a feeling it could be super obvious, but here goes.

"Consider the following regression equation:

Y = 50 +2A - 10B + 8C

1-What is the Y intercept?
2-What Y would you predict for a case where A=3, B=1, and C=0?"


I'm confused because I know the linear regression equation to be y = ax + b (or whatever variation you may use) with a being the gradient and b being the intercept in this case. So I'm not sure how this new equation fits into that structure - or if it's even meant to?

Thanks.

stapel
11-11-2015, 02:55 PM
"Consider the following regression equation:

Y = 50 +2A - 10B + 8C

1-What is the Y intercept?
How are A, B, and C defined? Are they just constants, so "50 + 2A - 10B + 8C" is just a number, and the equation is like "y = 0x + b = b"? ;)

Reboika
11-11-2015, 03:15 PM
How are A, B, and C defined? Are they just constants, so "50 + 2A - 10B + 8C" is just a number, and the equation is like "y = 0x + b = b"? ;)

Ohhh....that would make sense (I think). So this is more of an algebra question? With the "50 + 2A etc" representing the intercept, so the question is asking me to solve the equation?

stapel
11-11-2015, 03:20 PM
Ohhh....that would make sense (I think). So this is more of an algebra question? With the "50 + 2A etc" representing the intercept,...
If A, B, and C are defined (somewhere) as being constants then, yes, I think they're asking for the y-intercept of y = 0x + b, so the slope is m = 0 (the line is horizontal) and the y-intercept is at y = b.


so the question is asking me to solve the equation?
Under the stated assumption, I believe the second part of the question is asking you to plug the given values into the specified spots, and simplify in order to find the corresponding value of y. ;)

Reboika
11-11-2015, 03:29 PM
If A, B, and C are defined (somewhere) as being constants then, yes, I think they're asking for the y-intercept of y = 0x + b, so the slope is m = 0 (the line is horizontal) and the y-intercept is at y = b.


Under the stated assumption, I believe the second part of the question is asking you to plug the given values into the specified spots, and simplify in order to find the corresponding value of y. ;)


Hmm, it doesn't actually state anywhere - that is literally all the info I was given for that question. Although there was a regression question prior which asked for me to calculate a regression line from some (x,y) data - unrelated though I believe. But your answer made me realise that as that first equation stands, one can't really solve it just like that? So I think it could have been simply asking for a definition..! D'oh. That would then explain the nature of the second question - having given values for the constants afterwards. :)

Reboika
11-11-2015, 04:04 PM
If A, B, and C are defined (somewhere) as being constants then, yes, I think they're asking for the y-intercept of y = 0x + b, so the slope is m = 0 (the line is horizontal) and the y-intercept is at y = b.


Under the stated assumption, I believe the second part of the question is asking you to plug the given values into the specified spots, and simplify in order to find the corresponding value of y. ;)


Hmm, it doesn't actually state anywhere - that is literally all the info I was given for that question. Although there was a regression question prior which asked for me to calculate a regression line - unrelated though I think. But your answer made me realise that as that first equation stands, one can't really solve it just like that if I'm correct? So I think it could have been simply asking for a definition..! D'oh. That would then explain the nature of the second question - having given values for the constants afterwards. :)

Subhotosh Khan
11-11-2015, 06:59 PM
I am a graduate who is looking over past exam papers for statistics. Basically starting from scratch in my learning since I absorbed very little from when I did it in uni (was compulsory for psychology students to take statistics for two of the four years of the degree).
I have a question based on Regression, and I'm struggling to get it. See below. I have a feeling it could be super obvious, but here goes.

"Consider the following regression equation:

Y = 50 +2A - 10B + 8C

1-What is the Y intercept?
2-What Y would you predict for a case where A=3, B=1, and C=0?"


I'm confused because I know the linear regression equation to be y = ax + b (or whatever variation you may use) with a being the gradient and b being the intercept in this case. So I'm not sure how this new equation fits into that structure - or if it's even meant to?

Thanks.


I think

this is multi-variable linear regression - going towards ANOVA.

In that case, A, B & C are variables.

What is the definition of y-intercept?

cheddarMN
11-12-2015, 12:02 AM
Surely, a, b, and c are intended to be explanatory variables. So the Y intercept (when a, b, and c are zero) is 50.