机器学习题解（六）—— 异常探测与推荐系统

发表于 2021-10-29 更新于 2024-09-05 分类于题解，机器学习阅读次数：
本文字数： 885 阅读时长 ≈ 1 分钟

ex8.m

Part 2: Estimate the dataset statistics

Compute the mean of the data and the variances

1 2	mu = sum(X) / m; sigma2 = sum((X - repmat(mu, m, 1)).^2) / m; % X - 307 * 1 matrix

Part 3: Find Outliers

Compute the F1 score of choosing epsilon as the threshold and place the value in F1

% define the anomolous condition
    condition = pval < epsilon;
    % sum up to get each parameter
    fp = sum((condition == 1) & (yval == 0));
    fn = sum((condition == 0) & (yval == 1));
    tp = sum((condition == 1) & (yval == 1));
    % define precision and recall
    prec = tp / (tp + fp);
    rec = tp / (tp + fn);
    % use F1 score to estimate its performance
    F1 = (2 * prec * rec) / (prec + rec);

ex8_cofi.m

% compute cost function
J = sum(sum((((X * Theta' - Y) .* R) .^ 2))) / 2;
% compute gradient
X_grad = ((X * Theta' - Y) .* R) * Theta;
Theta_grad = ((X * Theta' - Y) .* R)' * X;
% compute cost function with regularization
J = J + lambda / 2 * ((sum(sum(Theta .^ 2)) + sum(sum(X .^ 2))));
% compute gradient with regularization
X_grad = X_grad + lambda * X;
Theta_grad = Theta_grad + lambda * Theta;