MyLinearFit.m
fbcb3eca7a6emaster

MyLinearFit.m
View Options

			classdef MyLinearFit < MyFit
			properties (Access=public)
			%Logical value that determines whether the data should be scaled or
			%not
			scale_data;
			end

			%Public methods
			methods (Access=public)
			%Constructor function
			function this=MyLinearFit(varargin)
			this@MyFit(...
			'fit_name','Linear',...
			'fit_function','a*x+b',...
			'fit_tex','$$ax+bx$$',...
			'fit_params',{'a','b'},...
			'fit_param_names',{'Gradient','Offset'},...
			varargin{:});
			end

			end

			methods (Access=protected)
			%Overload the doFit function to do polyFit instead of nonlinear
			%fitting. We here have the choice of whether to scale the data or
			%not.

			function doFit(this)
			if this.scale_data
			s_c=...
			polyfit(this.Data.scaled_x,this.Data.scaled_y,1);
			this.coeffs=convScaledToRealCoeffs(this,s_c);
			else
			%Fits polynomial of order 1
			this.coeffs=polyfit(this.Data.x,this.Data.y,1);
			end
			end
			end

			methods (Access=private)
			%Converts scaled coefficients to real coefficients
			function r_c=convScaledToRealCoeffs(this,s_c)
			[mean_x,std_x,mean_y,std_y]=calcZScore(this.Data);
			r_c(1)=std_y/std_x*s_c(1);
			r_c(2)=(s_c(2)-s_c(1)mean_x/std_x)std_y+mean_y;
			end

			function s_c=convRealToScaledCoeffs(this,r_c)
			[mean_x,std_x,mean_y,std_y]=calcZScore(this.Data);
			s_c(1)=std_x/std_y*r_c(1);
			s_c(2)=(r_c(2)-mean_y)/std_y+s_c(1)*mean_x/std_x;
			end
			end

			end