Find the relation of your plot to your theoretical analysis. In other words, if your theoretical analysis says that for a set of n points Q , the convex hull CH( Q ) O( g ( n )), does g ( n ) actually fit your empirical data? If so, what is the constant of proportionality k so that CH( Q ) = k g ( n )? If not, then which function g ( n ) best fits your empirical data, and what is the constant of proportionality? You can fit a function analytically using software or by trial and error in a spreadsheet, for example.