Cost function -비용함수/ 손실함수

출처: machine learning by andrew ng, stanford cousera lecture

수강일시: 2021.01.24 

 

www.coursera.org/learn/machine-learning/home

Coursera | Online Courses & Credentials From Top Educators. Join for Free | Coursera

 

in previeous lecture we tried to find out the housing price

cost function is way to measure difference in real value and estimated value.

so we can measure the accuracy of our hypothesis function by using a cost function.

 

 

cost function J

we want to minimize cost finction J to find a regression which can represent the values (or data) best

 

let's consider single parameter regression line(uni variable)

for value theta1, we can dot the J(theta1).

also, we can plot the J whatever the theta value is, and it will look like following graph in this problem:

so we can find the minimum value of J in this problem.

 

but how about multi variate problem?

 

It will look like this.

we can find our minimum value J* in 3dimension.

 

next time we will learn how to find this minimum value(by gradient discent)