plotting using a box plot

fafz1203

2016 10 18

0 个回答

6 次查看（30 天）

0 个投票

The following MATLAB code works correctly. After making this calculations I want to do 100 realizations of the training data (xTrain and yTrain). By that I mean, basically run it 100 times with different xTrain and yTrain. After that I want to plot these values I got on a box plot. Please see the attached image below to illustrate the box plot I want. Can anyone help with this.

Let me know if anything is unclear.

sigma = eye(10);
mu = repelem(zeros,10);
sd1 = 0.01; sd2 = 0.1; sd3 = 0.25;
xTrain = mvnrnd(mu,sigma,1000);
xTest = mvnrnd(mu,sigma,1000);
betaTruth = (mvnrnd (mu,sigma))';
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
epTest1 = noise(sd1); epTest2 = noise(sd2); epTest3 = noise(sd3);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
yTest1 = yValues(xTest,betaTruth,epTest1);
yTest2 = yValues(xTest,betaTruth,epTest2);
yTest3 = yValues(xTest,betaTruth,epTest3);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
betaHat1 = beta(xTrain,yTrain1);
betaHat2 = beta(xTrain,yTrain2);
betaHat3 = beta(xTrain,yTrain3);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
testError1 = error(xTest,betaHat1,yTest1);
testError2 = error(xTest,betaHat2,yTest2);
testError3 = error(xTest,betaHat3,yTest3);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function n = noise(x)
n = x.*randn(1000,1);
end
function e = error(x,b,y)
diff = ((x*b) - y).^2;
e = sum (diff(:))/1000;
end
function y = yValues(x,b,n)
y = (x*b) + n;
end
function b = beta(x,y)
b = (inv(x'*x)*x')*y;
end