Machine Learning by Andrew Ng - 1.4

Model Representation

In supervised learning, we have a data set and this data set is called a training set. The algorithm will learn from this data set and find pattern.

Training Set Of Housing Prices (Portland, OR).png

Notation:

• m = Number of training examples; 训练样本的数目
• x's = 'input' variables / features; 特征量
• y's = 'output' variables / 'target' variables; 目标变量
  (x, y) : one training example; 一个训练样本
  ( x ⁽ⁱ⁾, y ⁽ⁱ⁾) : i^th training example; i不是指数，而是指第i行的样本

How supervised learning algorithm works

image.png

1. Feed training set to learning algorithm.
2. Learning algorithm output a function h(hypothesis).
3. The function hypothesis takes in the value of x and tries to output the estimated value y.

• h is a function that maps from x's to y's.

Univariate linear regression : 一元线性回归
(Univeriate is a fancy way of saying one variable)
In this case, x is the only variable.