Regression (Regression)
"Return to moderate"
Rate Forecast:
- Regression (regression analysis) is used to establish how the correlation equation between the two or more simulation variables
- Predicted variable named: dependent variable (dependent variable), the output (output)
- It is used to predict variable called: argument (independent variable), the input (input)
- Linear regression comprises a variable and a dependent variable from
- With a relationship between two or more variables to simulate a straight line
- If contains two or more independent variables, multiple regression analysis (multiple regression) is referred to as
\[ a_\theta(x) = \theta_0 + \theta_1x \]
This equation corresponds to a straight line image, called the regression line. Wherein, \ (\ theta_1 \) is the slope of the regression line, \ (\ theta_0 \) is the intercept of the regression line. \ (X \) is because there is only one argument.
Linear regression - a positive correlation:
Linear regression - a negative correlation:
Linear regression - not relevant:
Solving equation coefficients:
It determines which line is preferably:
The cost function (Cost Function)
- Least squares method
- True value y, the predicted value \ (H_ \ Theta (X) \) , the squared error \ ((y -h_ \ theta ( x)) ^ 2 \)
- Find the parameters depending on the core, so that the sum of squared errors:
\[ j(\theta_0,\theta_1) = \frac{1}{2m}\sum_(i = 1)^m(y^i - h_\theta(x^i))^2 最小 \]