Calculate Sum of Square Error
195 views (last 30 days)
Show older comments
% Extract Data from Excel sheet to Metrix
Metrix = xlsread('D:\data.xlsx');
% X represent annual franchise fee and Y represent start up cost ($1000) for a pizza franchise
X = Metrix(:,1);
Y = Metrix(:,2);
% as we know regression line eq is ---> y = wx+b where w is slope and b is y-intercept
SUMxy = sum(X.*Y);
SUMx = sum(X);
SUMy = sum(Y);
n = length(X);
SUMx2 = sum(X.*X);
SUMxthen2 = SUMx*SUMx;
slope_W = (((n)*(SUMxy)) - (SUMx*SUMy))/((n*SUMx2)-(SUMxthen2));
YIntercept_B = (SUMy/n)-(slope_W*(SUMx/n));
x=linspace(0,2000);
eq_y = slope_W*x+YIntercept_B;
scatter(X,Y,'*');
hold on;
plot(x,eq_y);
hold off;
as we know SSE = (y-y_bar)^2 .. but i have not a y_bar values how to i extract y_bar values in Matrix ?
4 Comments
Accepted Answer
Wayne King
on 28 Oct 2013
Edited: Wayne King
on 28 Oct 2013
You get the yhat values by plugging your observed X-values into the equation determined by your regression fit. You have an equation that comes from estimating the slope and the intercept.
yhat = slope*x+intercept
You have a vector of X-values, the values you used in determining the model fit in the first place. So you just plug that vector in your fitted equation and you will get a vector of yhat values equal in length to your observations, then just use the code I gave you above.
2 Comments
More Answers (1)
Wayne King
on 28 Oct 2013
The sum of square error in regression is the 2-norm squared of the residuals, so if yhat are your fitted values, and y are the original observations, then
r = y-yhat;
SSE = norm(r,2)^2;
3 Comments
Wayne King
on 28 Oct 2013
yhat are your fitted values. You have a regression equation, it's your eq_y the values of that equation are your predicted values
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!