andrew ng的machine learning第一次作业

来源：互联网发布：医学图像处理算法编辑：程序博客网时间：2024/06/08 07:10

从今天开始，尝试着写csdn吧

今天主要学习了ng的第一次作业，我对这些machine learning的知识也不熟，一边做，一边看别人的解答，自己逐渐摸索，在这里做一个记录。

computeCost.m

product = 0;
tmp = 0;
summation = 0;
count = m;

%fprintf('m is %d \n', m);

for i=1:count
tmp = (X(i,:) * theta) - y(i,:);
result = tmp * tmp;
summation = summation + result;
%fprintf('i is %d \n', i);
end

J = (1 / (2*m)) * summation;

-----------------------------------------------------

gradientDescent.m

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

------------------------------------------------------

featureNormalize.m

%mu(1) = mean(X(:,1));
%mu(2) = mean(X(:,2));
mu = mean(X,1);

%sigma(1) = std(X(:,1));
%sigma(2) = std(X(:,2));
sigma = std(X);

for i = 1:size(X,2)
X_norm(:,i) = (X(:,i) - mu(:,i)) / sigma(:,i);
i = i + 1;
end

------------------------------------------------------

normalEqn.m

theta = inv(X'*X) * X' * y

------------------------------------------------------

查阅了这篇文章里的解答，这篇文章写的比我好。优秀的人真多，祖国有希望啊！

http://blog.csdn.net/yunlong34574/article/details/8869108

0 0