Question 156170: Hi, I need help
.
This is linear regression
.
I have these two lists
.
Let's say the left side is
.
Still heartbeat rate ! Racing heartbeat rate
... 50.......................... 166
... 60.......................... 168
... 70.......................... 170
... 80.......................... 172
.
I was told to use the first "x" as 0, and base the others on "0"
(For example, this is how you would write it in the calculator: 50 = 0, 60 = 10, 70 = 20, 80 = 30
.
If you put this in a calculator, using (0,10,20,30)as L1, (166,168,170,172) as L2
.
I get the regression line formula, of  It has told me to predict
.
Still heartbeat rate ! Racing heartbeat rate
... 40.......................... ???
... 65.......................... ???
... 84.......................... ???
.
If I put in the numbers ( -10, 15, 34)(based on the first "50" = 0
.
Using the formula
.
(-10) = =  =  = 
.
The other solutions are ( 169, and 172.8 )
.
Now, I found that if you just use the numbers themselves (50, 60, 70, 80) Instead of (0,10,20,30)
.
You come up with a different equation
.
The new equation =  ( The book I got this problem from, has this answer/equation)
.
If I predict the same heart rates ( 40, 65, 84)
.
If I use the new equation 
.
(40) = =  =  =
.
I get the same predictions, so which regression line is correct
.
(A)
.
(B)
.
Is it "A" or "B", my question is, "How do you know which regression line formula is correct?" It can be an infinite number of equations, depending on what you use for "L1"
.
Do you only use the First Number in list as "0" if it is a year?
.
Thanks ahead of time, Levi
Answer by stanbon(75887)   (Show Source):
You can put this solution on YOUR website! Using the orinal set of numbers gives you the equation you want:
y = 0.2x + 156
-------------------
Making the 50 becomes 0 change sinply
moved the regression line 10 to the left.
The y-value for 50 in the 1st set and 0
in the 2nd set are the same because you
only translated the x values.
--------------------------------
Each of your sets of x/y pairs gives rise
to a different regression line. I don't
know who told you to traslate 50 to 0;
you might ask them, why?
==========================
Cheers,
Stan H.
|
|
|
