Exercise 2: Linear Regression
话说LaTex用起来好爽
Matlab代码
迭代并且画出拟合曲线
Linear regression 公式如下
hθ(x)=θTx=∑i=0nθixi
(i是代表x的个数)
batch gradient descent update rule
θj:=θj−α1m∑i=1m(h(i)θ−y(i))x(i)j(for all j)
α=0.07
x = load('L:\\MachineLearning2016\\ex2x.dat');
y = load('L:\\MachineLearning2016\\ex2y.dat');
m = length(y);
x = [ones(m, 1), x];
theta=[0,0];%row vector
figure % open a new figure window
plot(x(:,2), y, 'o');
ylabel('Height in meters')
xlabel('Age in years')
hold on % Plot new data without clearing old plot
plot(x(:,2), x*transpose(theta), '-')
legend('Training data', 'Linear regression')
%迭代方式1
newTheta1 =theta(1,1) - transpose(x(:,1)) * (x*transpose(theta) -y) *0.07 * 0.02;
newTheta2 =theta(1,2) - transpose(x(:,2)) * (x*transpose(theta) -y) *0.07 * 0.02;
theta=[newTheta1,newTheta2];
%迭代方式2
for ii = 1:1500
theta = theta - transpose( x*transpose(theta) -y ) * x * 0.07 * 0.02;
if rem(ii,100) == 0
hold on % Plot new data without clearing old plot
plot(x(:,2), x*transpose(theta), '-')
end
end
hold on % 打印最后一条拟合曲线
plot(x(:,2), x*transpose(theta), '+')
画出J(θ)的图像
Understanding J(θ)
J(θ)=12m∗∑i=1m(h(i)θ−y(i))2(i means the ith of sample)
J_vals = zeros(100, 100); % initialize Jvals to 100x100 matrix of 0's
theta0_vals = linspace(-3, 3, 100);
theta1_vals = linspace(-1, 1, 100);
for i = 1:length(theta0_vals)
for j = 1:length(theta1_vals)
t = [theta0_vals(i); theta1_vals(j)];%column vector
J_vals(i,j) = sum( (x*t' -y).^2 ) * 0.01;
end
end
% Plot the surface plot
% Because of the way meshgrids work in the surf command, we need to
% transpose J_vals before calling surf, or else the axes will be flipped
J_vals = J_vals';
figure;
surf(theta0_vals, theta1_vals, J_vals)
xlabel('\theta_0'); ylabel('\theta_1')