Learning Log---Linear regression and logistic regression

Guide says:

All belong to the generalized linear regression category.

Linear regression analysis: Unary linearity (when a dependent variable, once, is a straight line);

Multivariate linear (the dependent variable has many, but also once, in the space is a plane);

Generalized linearity (high-dimensional linear regression, i.e., a hyper-plane)

(All at once, so called linear regression)

Nonlinear regression analysis: Not once, is a curve, some can be processed by linear model, called generalized linear model, such as logistic back

return;

Difficulty: Filter variables (dimensionality reduction techniques), avoid multiple collinearity (one variable depends on several other variables), observe fitting equations, evade

Avoid fitting ...

Use linear regression basis:

For linear regression, you can use correlation coefficients to determine whether these variables are suitable for linear fitting:

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M01/72/74/wKioL1XkFUiBpi6pAABz2PuWzro182.jpg "title=" Picture 1.png "alt=" wkiol1xkfuibpi6paabz2puwzro182.jpg "/> can prove its value between 1 and 1 through Cauchy inequality.

The closer you are to 1 or 1, the more linear the degree of correlation is. Positive correlation and negative correlation.

Learning Log---Linear regression and logistic regression