Angrew Machine Learning ex1

warmUpExercise.m

 

A = eye(5);

 

computeCost.m / computeCostMulti.m

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

gradientDescent.m / gradientDescentMulti.m

delta = X' * (X * theta - y); %the derivative part
delta = delta * alpha / m;theta = theta - delta;

 

featureNormalize.m

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

normalEqn.m

theta = pinv(X' * X) * X' * y; % We tack this measure rather than pinv(X) * y in order to prevent the possiblity that X is non-invertible

 

ex1_multi.m

price = [1 (1650-mu(1))/sigma(1) (3-mu(2))/sigma(2)] * theta; %Gradient Descent

price = [1 1650 3] * theta; % Normal Equation

这是笔者自己想到一些对吴恩达机器学习课程的编程作业的实现方式，如果你有更好的实现方式，欢迎在评论区讨论。

这里只是部分代码，全部代码在 

https://download.csdn.net/download/ti_an_di/10590380