Great God Notes http://scuel.gitee.io/ml-andrewng-notes/week2.html
The following figures or content are added for ease of understanding
4.1 Multiple Features:
4.2 Multivariate Gradient Descent
Ichiyota:
4.3 Gradient Descent - Eigenvalue Scaling
4.4 Gradient Descent - Learning Efficiency
4.5 Features and Polynomial Regression
We can change the behavior or curve of our hypothetical function by making it a quadratic, cubic or square root function (or any other form). Linear regression can only fit data with a straight line, and sometimes it is necessary to use a curve to fit the data, that is, polynomial regression (Polynomial Regression) . When using polynomial regression, keep in mind that feature scaling is very necessary, such as The range is 1-1000, then The range is 1-1000000. If feature scaling is not applied, the range is more inconsistent and more likely to affect efficiency.
4.6 Normal equation (normalized equation):
The derivation formula of the matrix is the key, the picture is reproduced from https://blog.csdn.net/perfect_accepted/article/details/78383434
4.7 Irreversible Normal Equations
Selected Lectures