Re: 'trend value' formula needed


khany schrieb:

Hi all,

hope this isn't too off-topic but i could really do with some help

i have, say, at least 5 records with 3 columns of data based on
industry activity (so
not pre-determined by me)

the values of the first 2 columns roughly reflect the value of the
third.

i want to be able to pass in 2 arbitrary values for the first 2 columns
and get a trend/predicted value for the third back.

eg. some records I already have:

A:22000, B:24, C: 5000
A:36000, B:35, C: 3200
A:56000, B:43, C: 1800

so as A and B increase (not directly related) C decreases.

So if I pass in A:39000 and B:37 what would C be? more importantly,
whats the formula?

thanks in advance

khany

How many records do you have in total?
With enough records (and a not too "cloudy" distribution) you might
guess a good /type/ of dependency of C (linear? reciprocal of linear?
....)
Is there any theory behind that which say that there should exist some
approximate relation of a certain kind between A, B and C?
Otherwise, what prevents the occurrence of records
A: 20000, B: 20, C: 0
A: 20000, B: 50, C: 0
A: 40000, B: 20, C: 0
A: 40000, B: 50, C: 0
A: 20000, B: 20, C: 1000
A: 20000, B: 50, C: 1000
A: 40000, B: 20, C: 1000
A: 40000, B: 50, C: 1000 ?

.