I am trying to adjust an equation to account for a non-linear trend, while preserving the final product without a transformation.
My target, is to estimate a concentration gradient, e.g. If the air has x gas concentration, then a partially-closed box (z) has 'z = x + y' gas concentration. y is always positive and higher than 0. But I would like to refine the + y to make it non-linear. My current calculation for + y is: (100-x)/2, where x is on a scale of 0-100 where 100 is the maximum concentration the gas can be.
Currently, this adjustment is almost right for the data I am comparing it to. However, '/2' should be replaced with something else, because it overestimates y at low x and underestimates y at high x.
Do you know how I should make this adjustment?