W2编程作业——Linear Regression

必做

warmUpExercise.m

A = eye(5);

plotData.m

plot(x,y,'rx');
axis([4,24,-5,25]);
ylabel('Profit in $10,000s');
xlabel('Population of City in 10,000s');

computeCost.m

J=(1/(2*m))*sum((X*theta-y).^2);

gradientDscent.m

theta=theta-(alpha/m)*(X'*(X*theta-y));

选做

featureNormalize.m

mu=mean(X);
sigma=std(X);
X_norm=(X_norm-mu)./sigma;

computeCostMulti.m

J=1/(2*m)*(X*theta-y)'*(X*theta-y);

gradientDexcentMulti.m

theta=theta-(alpha/m)*(X'*(X*theta-y));

normalEqn

theta=(pinv(X'*X))*X'*y;

ex1_multi.m修改部分

Part 2: Gradient Descent
补充对比不同学习速率alpha

alpha2 = 0.0001;
alpha3 = 1;
theta2= zeros(3, 1);
theta3= zeros(3, 1);
[theta2, J_history2] = gradientDescentMulti(X, y, theta2, alpha2, num_iters);
[theta3, J_history3] = gradientDescentMulti(X, y, theta3, alpha3, num_iters);
figure;
plot(1:numel(J_history), J_history,'b','LineWidth', 2);
title("alpha=0.01,0.0001,1对比");
hold on;
plot(1:numel(J_history2), J_history2,'r','LineWidth', 2);
hold on;
plot(1:numel(J_history3), J_history3,'k','LineWidth', 2);
legend('alpha=0.01','alpha=0.0001','alpha=1')
xlabel('Number of iterations');
ylabel('Cost J');

预测值

predict_x=[1650,3];
predict_x=(predict_x-mu)./sigma;
price = [1,predict_x]*theta; % You should change this

Part 3: Normal Equations
预测值

price = [1,1650,3]*theta; % You should change this

提交作业

运行submit.m后
输入coursera注册时的邮箱
输入作业提交页面生成的识别码
【如果是下载的以前的作业，需要重新下载 submit.m 以及 lib/submitWithConfiguration.m】