supervised learning

supervised learning

标签（空格分隔）： ML supervised-learning

目录

• supervised learning
  • 1. predicate the housing price
  • 2.one key point
  • 3.no cap on the number of features of each entries in dataset(SVM)
  • 4.main principle behind supervised learning

---

I have never encountered any machine learning before,so it is suitable to start with examples.

1. predicate the housing price

axis explanation:

\[\begin{cases} horizontal \quad axis \longmapsto the \quad price \quad of\quad house\quad in\quad thousands\quad of\quad dollar\\ vertical \quad axis \longmapsto the \quad size \quad of \quad house \quad in \quad square \quad feets\\ \end{cases} \]

An assumption:

Given two unknowns, how can we fit these data point on the 2-D plane?

\(\implies \quad\)\( \begin{cases} 1.fit \quad a \quad straight \quad line \quad to \quad the \quad data\\ 2.fit \quad a \quad quadratic \quad function \quad or \quad second-order \quad polynomial\\ \end{cases} \)

2.one key point

As we can see the results shown above,it's continuous rather than discrete like 0,1.Generally speaking,the categories of supervised learning usually depends on result:

• classification (dicrete result)
• regression (continuous result)

3.no cap on the number of features of each entries in dataset(SVM)

In some supervised learning model,not only two or three features but also up to thousands of features are involved with supervised learning algorithm,which can be accomplished by supported vector machine.

4.main principle behind supervised learning

In each example in our data set,we're told what is the correct answer in order to measure how high accuracy our supervised learning algorithm can reach.
(important!)the measurement for machine learning model accounts for a big proportion of this course.