# Matlab Linear Fitting Function

## Linear fit

Copy this fitting route, from the Physics 111-Lab Library Site, to your My Documents analysis folder then use them to fit your data while using MatLab. These scripts should be in the directory folder where you are using Matlab. Linear Fit file %Load this into Matlab to excute function [ outStruct ] = linfit( x, y, dy ) %LINFIT Performs a Linear Fit on data and calculates % uncertainty in fits. Fit is y = A + B*x % % Part of the Physics 111 MATLAB Fitting Toolkit - 2009 % % INPUTS: x, y, (dy) % All inputs must be the same size and either Nx1 or 1xN in dimension. % If dy is not provided, the fit will assume an equal weights fitting % % RETURNS: A struct containing the following fields: % A: Y-intercept % B: Slope % siga/sigb: Uncertainties in A/B respectively % fitx/fity: The x/y coordinates for the fit. X coordinates % are necessarily the same as the input. % chi2: A chi^2 estimate of goodness of fit. NOT reduced. % For degrees of freedom for a reduced chi2, use the form % of the equation along with the size of the fitx/y arrays.

N = length(x); x = x(:); y = y(:); if (~exist('dy','var'))

delta = (N*sum(x.^2)-((sum(x))^2)); A = ((sum(x.^2)*sum(y))-(sum(x)*sum(x.*y)))/delta; B = (N*sum(x.*y)-sum(x)*sum(y))/delta; if N > 2 sigy = sqrt((1/(N-2))*sum((y - A - B*x).^2)); end siga = sigy*(sqrt(sum(x.^2)/delta)); sigb = sigy*(sqrt(N/delta));

else

dy = dy(:); weights = 1./(dy.^2); delta = (sum(weights)*sum(weights.*(x.^2))) - ... (sum(weights.*x))^2; A = ((sum(weights.*(x.^2))*sum(weights.*y)) - ... (sum(weights.*x)*sum(weights.*x.*y))) / delta; B = ((sum(weights)*sum(weights.*x.*y)) - ... (sum(weights.*x)*sum(weights.*y))) / delta; siga = sqrt(sum(weights.*(x.^2))/delta); sigb = sqrt(sum(weights)/delta);

end

e_y = (A + B*x); chi2 = sum(((y - e_y).^2)./e_y);

outStruct = struct(); outStruct.A = A; outStruct.B = B; outStruct.siga = siga; outStruct.sigb = sigb; outStruct.fitx = x; outStruct.fity = e_y; outStruct.chi2 = chi2; end