Shulou(Shulou.com)06/01 Report--

What does the integrated learning of R language classification algorithm refer to? I believe that many inexperienced people are at a loss about it. Therefore, this paper summarizes the causes and solutions of the problem. Through this article, I hope you can solve this problem.

1. Principle analysis of integrated learning (Bootstrap Aggregating):

Bagging is the abbreviation of Bootstrap Aggregating. To put it simply, several different training sets are obtained by using boostrap sampling, and models are established based on these training sets, that is, a series of base classifiers are obtained. because these classifiers come from different training samples, their prediction effects on the same test set are different. Therefore, the Bagging algorithm then votes (classification problem) and averages (regression problem) on a series of prediction results of the base classifier, so as to get the final prediction result of each test set sample. the integrated results are often accurate and stable.

For example, if the current base classifier 1 to 10 and their prediction results for a sample are class 1, 2, 1, and 1, respectively, then the final result given by Bagging is "the sample belongs to category 1," because most base classifiers vote for 1.

AdaBoost (Adaptive Boosting) is more ingenious than Bagging algorithm, and generally speaking it is a better integrated classification algorithm, especially in the case of uneven distribution of data sets, its advantage is more significant. This algorithm is put forward before Bagging, but it is higher than Bagging in complexity and effect.

AdaBoost is also an integrated algorithm based on several base classifiers, but it is different from the simple synthesis of a series of prediction results by Bagging. In the process of constructing the base classifier in turn, the algorithm will adjust the probability of each sample being selected according to the prediction results of the last base classifier to each training set. Specifically, if the sample I is misclassified in the prediction of the previous base classifier, then in the process of training sample extraction of the base classifier, the sample I will be given a higher weight. so that it can be selected with a greater possibility, so as to improve the probability of being correctly classified.

Such a process of real-time weight adjustment is the advantage of AdaBoost algorithm. By integrating several complementary base classifiers, it significantly improves the stability and accuracy of the integrated classifier. In addition, the base classifier selection of Bagging and AdaBoost is arbitrary, but most of us use the decision tree, because the decision tree can deal with various types of variables such as numerical value / category / order at the same time, and the selection of variables is easy.

two。 Application in R language

The ensemble learning algorithm is mainly applied to the bossting function in adabag package.

Bossting (formula,data,boos=TRUE,mfinal=100,coeflearn='Breiman',control)

3. Taking iris dataset as an example for analysis

1) apply the model and view the corresponding parameters of the model

Fit_bag=bagging (Species~.,data_train,mfinal=5,control=rpart.control (maxdepth=3))

Fit_ bag [1: length (bag)]

2) Predictive analysis

Pre_bag=predict (fit_bag,data_test)

After reading the above, have you mastered what the integrated learning of R language classification algorithm refers to? If you want to learn more skills or want to know more about it, you are welcome to follow the industry information channel, thank you for reading!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.