lr-Logistic regression

Because logistic regression is very important for calculating advertising. is also our usual advertising recommendations, CTR estimates the most commonly used algorithm. So write a separate article to discuss.

Refer to this article: http://www.cnblogs.com/sparkwen/p/3441197.html

Logistic regression is only based on the linear regression, the application of a logical function , but also because of this logic function, logical regression became a machine learning field of a dazzling star, is the core of computational advertising .

Logistic regression is only based on the linear regression, the application of a logical function, but also because of this logic function, logical regression became a machine learning field of a dazzling star, is the core of computational advertising.

In the industry, the LR model is very popular, mainly because the LR model is a logarithmic linear model, the implementation is simple, easy to parallel, large-scale expansion is convenient, iterative speed, and the use of the characteristics of a better interpretation of the predicted output between 0 and 1 fit probability model. However, the linear model lacks the accurate characterization to the nonlinear relation, the characteristic combination can add the nonlinear expression and enhance the expression ability of the model. In addition, in AD LR, the basic features can be considered for global modeling, the combination of features more granular, is personalized modeling, because in this large-scale discrete LR, the single-to-global modeling will be partial users, modeling and data for each user is not easy to fit the model number explosion, so the basic features + Combination features are both global and personalized.

The following is "machine learning"-Zhou Zhihua's reading notes:

P54

P58 Generalized linear regression

P58 finally talked about logistic regression. is actually a logistic Regression. So the book insists on translating the regression into a rate of return. Pedantic. Make people almost do not understand.

Advantages:

Solution:

About the likelihood function, like the following explanation:

In statistics, the likelihood function (likelihood functions), or simply likelihood, is a function of statistical model parameters. Given output x, the likelihood function L (θ|x) on the parameter θ is equal to the probability of the variable x after the given parameter θ: L (θ|x) =p (x=x|θ). Likelihood functions play an important role in inferential statistics (statistical inference) Especially in the parameter estimation method. In textbooks, likelihood is often used as a synonym for "probability". But in statistics, they have different uses. The probability describes the output of a random variable when a known parameter is used, and the possible value of an unknown parameter when describing the output of a known random variable. For example, in the case of "throwing 10 times on a positive and negative coin", we can ask how much the "probability" of 10 times the coin landed was positive, and for "10 times on a coin, the landing is positive", we can ask, "What is the likelihood" of the coin's positive and negative face?

P60 3.4 Linear discriminant analysis (Linear discriminant analyses, LDA)

Because LR is extremely important for computing advertising. Also to add.

lr-Logistic regression