Linear Regression - Suspected Typo?

Question

anonymous · Answer

Posted this to StackExchange but got no answers but I'm hoping someone here will be able to take a look: http://math.stackexchange.com/questions/1561458/linear-regression-questions-suspected-typo Replying here is fine, I just didn't want to retype the whole lot.

amistre64 · Answer

the slope of the line from an online calculator for the regression is:

-205.29046991994

which is what you got i believe ... so id go with typo.

anonymous · Answer

Thanks, just checking I wasn't going mad. Any idea about the last question?

amistre64 · Answer

youre 'corrections' were not complete

amistre64 · Answer

consider the point slope form of the linear regression line:

y = m(x-avgX) + avgY

such that the y intercept is:  -m*avgX + avgY

i believe you adjusted the value of m, but did not consider how it affects the overall line outcome.

amistre64 · Answer

when i correct for it, using their typoed slope 

4836.2623-x/(.004781)=4900

... i get x=-.30473

anonymous · Answer

Oh thank you, that explains it. Also, someone replied to me on StackExchange explaining that it wasn't at typo and was due to the fact they modelled it differently.

amistre64 · Answer

ah, i was wondering a little why you did 1/m for the ELR.  I spose we should have had a better definition of what ELR ... temp with respect to altitude would have given us a function whose independant variable was altitude, not temperature.