Fit data into a circle

Specific Domains Statistics
, ,
9

This is not the cleanest code in the world (some allocations that could be trimmed at least), and seems a bit fragile with respect to initial conditions, but here’s a working example using LsqFit.

using LsqFit
	
data = [(-2.0,1.0), (0.0, 3.0), (2.0, 1.0)]
	
ptrs = 1:length(data)
	
y = zeros(length(data))
	
function circle_model(t,p)
	out = []
	x0,y0,r = p
	for ptr ∈ ptrs
		x,y = data[ptr]
		push!(out,(x-x0)^2+(y-y0)^2-r^2)
	end
	out
end
	
p0 = [0.0, 0.0, 1.0] # using 0.0 as initial radius fails
	
fit = curve_fit(circle_model, ptrs, y, p0)
	
fit.param
1 Like