Machine Learning Day No. 0
Welcome reprint, please indicate the source (Http://blog.csdn.net/tonyshengtan), respect for labor, respect for knowledge, welcome to discuss.
The opening crap.
Back to write a blog, although always know that learning is not the end, but still will doubt, learn to what extent can find a job like this (spit groove: The work is too disgusting, the daily task is to sing the praises, whitewash, shirk responsibility, like to do technology students do not come to those so-called treatment, stability of the celestial enterprise , ruined life).
Needless to say, the choice of their own, will do.
Linear regression (Linear Regression)
First of all, what is regression, regression is called regression analysis, is a statistical analysis method used to determine the two or more variables of the interdependence of a relationship.
Popular points, such as the following picture, from Stanford public class "Machine learning":
The horizontal axis represents the variable housing area, the ordinate represents the house price, through the image data can presumably speculate, there is a dependency between them, or through common sense also know that the larger the price of the house, and the goal of linear regression is to find such a "line" (dependency expression) to fit the data.
Next summarize the meaning of the downline, the line must be straight line, but not necessarily a two-dimensional straight line, can be multidimensional straight line, can not be the other shape of the line (curved), but also need to distinguish between the point and polygon.
Blog structure
The content of the introduction of linear regression includes the following (there may be adjustments, this article continues to update):
- Least squares
- Minimum Average method
- A review of probability knowledge
- Maximum Likelihood method
Local Weighting method
SummarizeThe article is probably structured so, continuously updated, welcome to pay attention to
Machine learning--a survey of linear regression