function gradient = lsqGradient(parameter, x, y)
% The gradient of the least square error
%
% Arguments: parameter, vector containing slope and intercept 
%                       as the 1st and 2nd element 
%            x, vector of the input values
%            y, vector of the corresponding measured output values
%
% Returns: the gradient as a vector with two elements
  h = 1e-6;   % stepsize for derivatives
  partial_m = (lsqError([parameter(1)+h, parameter(2)], x, y) - lsqError(parameter, x, y))/ h;
  partial_n = (lsqError([parameter(1), parameter(2)+h], x, y) - lsqError(parameter, x, y))/ h;
  gradient = [partial_m, partial_n];
end