Lineary regression with 3 datasets

  • Thread starter Thread starter PS
  • Start date Start date
P

PS

Hi,

Maybe someone can help me...
Ive got 3 sets of data.

x y z
4 6 8
5 7 8
3 4 10

Can a function predict the next z if i give it a set of x og y´s ?
F.inst

5 7 ?

Best regards,

Sauer
 
Yes, use LINEST. Help for LINEST is misleading, it fits models that are
linear in the unknowns, not just simple straight lines. If your three
columns are in A1:C3, then select 3 blank adjacent cells in a row and
array enter (Ctrl-Shift-Enter)
=LINEST(C1:C3,A1:B3)

to get -2, 2, 12; which means that the estimated relationship is
z = 12 + 2*x - 2*y
which exactly fits your three observations since you have solved three
equations in three unknowns. You would need additional observations
from the unknown relationship to determine whether this is really a good
fit or whether you are just chasing noise.

Jerry
 
Thx a lot.

Best regards,

Peter,

Jerry W. Lewis said:
Yes, use LINEST. Help for LINEST is misleading, it fits models that are
linear in the unknowns, not just simple straight lines. If your three
columns are in A1:C3, then select 3 blank adjacent cells in a row and
array enter (Ctrl-Shift-Enter)
=LINEST(C1:C3,A1:B3)

to get -2, 2, 12; which means that the estimated relationship is
z = 12 + 2*x - 2*y
which exactly fits your three observations since you have solved three
equations in three unknowns. You would need additional observations
from the unknown relationship to determine whether this is really a good
fit or whether you are just chasing noise.

Jerry
 
Back
Top