# How to distinguish the contribution of two variables in a large data set

I'm trying to solve a statistics problem with no formal training in the area, so please bear with me.

I have a function with variables of pounds of product and number of figures which together give a price. So, for example, my data might look like this:

100 pounds + 50 figures = $400 65 pounds + 25 figures =$300

70 pounds + 70 figures = $430 And so on. Now, my question is: how do I determine what effect each one of the variables has on the final price in my large data set? Obviously I want to end up with the highest price per figure per pound, but I'm not sure how to tackle the problem. Any tips would be appreciated! Also if I can help by clarifying the problem I'll do my best. - ## 1 Answer The first thing that comes to mind is linear regression. You'd have $$\text{constant} + a\cdot\text{pounds} + b\cdot\text{figures} + \text{error} = \text{\}$$ where$a$is in dollars per pound and$b$is dollars per figure, and the "constant" is of course simply in dollars, as is the "error". The two coefficients$a$and$b\$ and the "constant" would be estimated by least squares, i.e. the estimates would be the values of those variables that minimize the sum of squares of the "errors". And since they are estimates, the "errors" would be "residuals", i.e. observable estimates of "errors". There are lots of standard software packages that do this.

However, upon seeing your whole data set and putting it through a few algorithms, I could easily decide that a different approach is needed.

-