Louis R wrote:looks like what you have is a broken power law: the exponent changes value at least once.
plotting it on a [iirc] log-log scale will produce straight lines with slope = exponent - and show a knee where the exponent changes.
Jonathan_S wrote:Initial eyeball looks like the LACs are slightly off line. Might play with removing them.
If you look closely at the smaller ships, one can distinctly see at least two different equations at work. Seems to be three equations when broken into age of the ship in question. This will utterly destroy any curve fit. It would appear that a scaler component equating to compensator efficiency is at work. Likewise there is a limiting factor for "FAT" ships. BCP, CLAC, CARAVAN. ... I am wondering if the NIKE BCL is a mistake? Though if I label the BCP as a "FAT" ship it could work while removing older ships and only use more modern ships, Culverin, Roland, SAG-B/C etc, it fits.
One wedge/tonnage that does not fit is the Edward Saganami. It fits no curve. Remove it.
REGARDING "FAT" ships:
I would have expected the "FAT" ships, to require a larger wedge for their tonnage. (Less efficient) Instead we see that "FAT" ships have a smaller wedge for their tonnage in comparison... I wonder if someone forgot to invert the fraction...
PS> If you flip your graph with tonnage on the Y-axis you get a nice simple power equation curve... Make sure to break out the old ships from the new.