R语言中的机器学习

雨点打透心脏的1/2处 2024-02-17 13:01 50阅读 0赞

**转载自 R中文论坛(http://rbbs.biosino.org/Rbbs/posts/list/192.page)**

**Machine Learning & Statistical Learning （机器学习 & 统计学习）**

网址：[http://cran.r-project.org/web/views/MachineLearning.html][http_cran.r-project.org_web_views_MachineLearning.html]维护人员：Torsten Hothorn

版本：2008-02-18 18:19:21

翻译：R-fox, 2008-03-18

机器学习是计算机科学和统计学的边缘交叉领域，R关于机器学习的包主要包括以下几个方面：

**1）神经网络（Neural Networks）：**

nnet包执行单隐层前馈神经网络，nnet是VR包的一部分（[http://cran.r-project.org/web/packages/VR/index.html][http_cran.r-project.org_web_packages_VR_index.html]）。

**2）递归拆分（Recursive Partitioning）：**

递归拆分利用树形结构模型，来做回归、分类和生存分析，主要在rpart包（[http://cran.r-project.org/web/packages/rpart/index.html][http_cran.r-project.org_web_packages_rpart_index.html]）和tree包（[http://cran.r-project.org/web/packages/tree/index.html][http_cran.r-project.org_web_packages_tree_index.html]）里执行，尤其推荐rpart包。Weka里也有这样的递归拆分法，如：J4.8, C4.5, M5，包Rweka提供了R与Weka的函数的接口（[http://cran.r-project.org/web/packages/RWeka/index.html][http_cran.r-project.org_web_packages_RWeka_index.html]）。

party包提供两类递归拆分算法，能做到无偏的变量选择和停止标准：函数ctree()用非参条件推断法检测自变量和因变量的关系；而函数mob()能用来建立参数模型（[http://cran.r-project.org/web/packages/party/index.html][http_cran.r-project.org_web_packages_party_index.html]）。另外，party包里也提供二分支树和节点分布的可视化展示。

mvpart包是rpart的改进包，处理多元因变量的问题（[http://cran.r-project.org/web/packages/mvpart/index.html][http_cran.r-project.org_web_packages_mvpart_index.html]）。rpart.permutation包用置换法（permutation）评估树的有效性（[http://cran.r-project.org/web/packages/rpart.permutation/index.html][http_cran.r-project.org_web_packages_rpart.permutation_index.html]）。knnTree包建立一个分类树，每个叶子节点是一个knn分类器（[http://cran.r-project.org/web/packages/knnTree/index.html][http_cran.r-project.org_web_packages_knnTree_index.html]）。LogicReg包做逻辑回归分析，针对大多数自变量是二元变量的情况（[http://cran.r-project.org/web/packages/LogicReg/index.html][http_cran.r-project.org_web_packages_LogicReg_index.html]）。maptree包（[http://cran.r-project.org/web/packages/maptree/index.html][http_cran.r-project.org_web_packages_maptree_index.html]）和pinktoe包（[http://cran.r-project.org/web/packages/pinktoe/index.html][http_cran.r-project.org_web_packages_pinktoe_index.html]）提供树结构的可视化函数。

**3）随机森林（Random Forests）：**

randomForest 包提供了用随机森林做回归和分类的函数（[http://cran.r-project.org/web/packages/randomForest/index.html][http_cran.r-project.org_web_packages_randomForest_index.html]）。ipred包用bagging的思想做回归，分类和生存分析，组合多个模型（[http://cran.r-project.org/web/packages/ipred/index.html][http_cran.r-project.org_web_packages_ipred_index.html]）。party包也提供了基于条件推断树的随机森林法（[http://cran.r-project.org/web/packages/party/index.html][http_cran.r-project.org_web_packages_party_index.html]）。varSelRF包用随机森林法做变量选择（[http://cran.r-project.org/web/packages/varSelRF/index.html][http_cran.r-project.org_web_packages_varSelRF_index.html]）。

**4）Regularized and Shrinkage Methods：**  
lasso2包（[http://cran.r-project.org/web/packages/lasso2/index.html][http_cran.r-project.org_web_packages_lasso2_index.html]）和lars包（[http://cran.r-project.org/web/packages/lars/index.html][http_cran.r-project.org_web_packages_lars_index.html]）可以执行参数受到某些限制的回归模型。elasticnet包可计算所有的收缩参数（[http://cran.r-project.org/web/packages/elasticnet/index.html][http_cran.r-project.org_web_packages_elasticnet_index.html]）。glmpath包可以得到广义线性模型和COX模型的L1 regularization path（[http://cran.r-project.org/web/packages/glmpath/index.html][http_cran.r-project.org_web_packages_glmpath_index.html]）。penalized包执行lasso (L1) 和ridge (L2)惩罚回归模型（penalized regression models）（[http://cran.r-project.org/web/packages/penalized/index.html][http_cran.r-project.org_web_packages_penalized_index.html]）。pamr包执行缩小重心分类法(shrunken centroids classifier)（[http://cran.r-project.org/web/packages/pamr/index.html][http_cran.r-project.org_web_packages_pamr_index.html]）。earth包可做多元自适应样条回归（multivariate adaptive regression splines）（[http://cran.r-project.org/web/packages/earth/index.html][http_cran.r-project.org_web_packages_earth_index.html]）。

**5）Boosting :**

gbm包（[http://cran.r-project.org/web/packages/gbm/index.html][http_cran.r-project.org_web_packages_gbm_index.html]）和boost包（[http://cran.r-project.org/web/packages/boost/index.html][http_cran.r-project.org_web_packages_boost_index.html]）执行多种多样的梯度boosting算法，gbm包做基于树的梯度下降boosting，boost包包括LogitBoost和L2Boost。GAMMoost包提供基于boosting的广义相加模型(generalized additive models)的程序（[http://cran.r-project.org/web/packages/GAMMoost/index.html][http_cran.r-project.org_web_packages_GAMMoost_index.html]）。mboost包做基于模型的boosting（[http://cran.r-project.org/web/packages/mboost/index.html][http_cran.r-project.org_web_packages_mboost_index.html]）。

**6）支持向量机（Support Vector Machines）：**

e1071包的svm()函数提供R和LIBSVM的接口 （[http://cran.r-project.org/web/packages/e1071/index.html][http_cran.r-project.org_web_packages_e1071_index.html]）。kernlab包为基于核函数的学习方法提供了一个灵活的框架，包括SVM、RVM……([http://cran.r-project.org/web/packages/kernlab/index.html][http_cran.r-project.org_web_packages_kernlab_index.html]) 。klaR 包提供了R和SVMlight的接口（[http://cran.r-project.org/web/packages/klaR/index.html][http_cran.r-project.org_web_packages_klaR_index.html]）。

**7）贝叶斯方法（Bayesian Methods）：**

BayesTree包执行Bayesian Additive Regression Trees (BART)算法（[http://cran.r-project.org/web/packages/BayesTree/index.html][http_cran.r-project.org_web_packages_BayesTree_index.html]，[http://www-stat.wharton.upenn.edu/~edgeorge/Research\_papers/BART%206–06.pdf][http_www-stat.wharton.upenn.edu_edgeorge_Research_papers_BART_206_06.pdf]）。tgp包做Bayesian半参数非线性回归（Bayesian nonstationary, semiparametric nonlinear regression）（[http://cran.r-project.org/web/packages/tgp/index.html][http_cran.r-project.org_web_packages_tgp_index.html]）。

**8）基于遗传算法的最优化（Optimization using Genetic Algorithms）：**  
gafit包（[http://cran.r-project.org/web/packages/gafit/index.html][http_cran.r-project.org_web_packages_gafit_index.html]）和rgenoud包（[http://cran.r-project.org/web/packages/rgenoud/index.html][http_cran.r-project.org_web_packages_rgenoud_index.html]）提供基于遗传算法的最优化程序。

**9）关联规则（Association Rules）：**  
arules包提供了有效处理稀疏二元数据的数据结构，而且提供函数执Apriori和Eclat算法挖掘频繁项集、最大频繁项集、闭频繁项集和关联规则（[http://cran.r-project.org/web/packages/arules/index.html][http_cran.r-project.org_web_packages_arules_index.html]）。

**10）模型选择和确认（Model selection and validation）：**  
e1071包的tune()函数在指定的范围内选取合适的参数（[http://cran.r-project.org/web/packages/e1071/index.html][http_cran.r-project.org_web_packages_e1071_index.html]）。ipred包的errorest()函数用重抽样的方法（交叉验证，bootstrap）估计分类错误率（[http://cran.r-project.org/web/packages/ipred/index.html][http_cran.r-project.org_web_packages_ipred_index.html]）。svmpath包里的函数可用来选取支持向量机的cost参数C（[http://cran.r-project.org/web/packages/svmpath/index.html][http_cran.r-project.org_web_packages_svmpath_index.html]）。ROCR包提供了可视化分类器执行效果的函数，如画ROC曲线（[http://cran.r-project.org/web/packages/ROCR/index.html][http_cran.r-project.org_web_packages_ROCR_index.html]）。caret包供了各种建立预测模型的函数，包括参数选择和重要性量度（[http://cran.r-project.org/web/packages/caret/index.html][http_cran.r-project.org_web_packages_caret_index.html]）。caretLSF包（[http://cran.r-project.org/web/packages/caretLSF/index.html][http_cran.r-project.org_web_packages_caretLSF_index.html]）和caretNWS（[http://cran.r-project.org/web/packages/caretNWS/index.html][http_cran.r-project.org_web_packages_caretNWS_index.html]）包提供了与caret包类似的功能。

**11）统计学习基础（Elements of Statistical Learning）：**  
书《The Elements of Statistical Learning: Data Mining, Inference, and Prediction 》（[http://www-stat.stanford.edu/~tibs/ElemStatLearn/][http_www-stat.stanford.edu_tibs_ElemStatLearn]）里的数据集、函数、例子都被打包放在ElemStatLearn包里（[http://cran.r-project.org/web/packages/ElemStatLearn/index.html][http_cran.r-project.org_web_packages_ElemStatLearn_index.html]）。

[http_cran.r-project.org_web_views_MachineLearning.html]: http://cran.r-project.org/web/views/MachineLearning.html
[http_cran.r-project.org_web_packages_VR_index.html]: http://cran.r-project.org/web/packages/VR/index.html
[http_cran.r-project.org_web_packages_rpart_index.html]: http://cran.r-project.org/web/packages/rpart/index.html
[http_cran.r-project.org_web_packages_tree_index.html]: http://cran.r-project.org/web/packages/tree/index.html
[http_cran.r-project.org_web_packages_RWeka_index.html]: http://cran.r-project.org/web/packages/RWeka/index.html
[http_cran.r-project.org_web_packages_party_index.html]: http://cran.r-project.org/web/packages/party/index.html
[http_cran.r-project.org_web_packages_mvpart_index.html]: http://cran.r-project.org/web/packages/mvpart/index.html
[http_cran.r-project.org_web_packages_rpart.permutation_index.html]: http://cran.r-project.org/web/packages/rpart.permutation/index.html
[http_cran.r-project.org_web_packages_knnTree_index.html]: http://cran.r-project.org/web/packages/knnTree/index.html
[http_cran.r-project.org_web_packages_LogicReg_index.html]: http://cran.r-project.org/web/packages/LogicReg/index.html
[http_cran.r-project.org_web_packages_maptree_index.html]: http://cran.r-project.org/web/packages/maptree/index.html
[http_cran.r-project.org_web_packages_pinktoe_index.html]: http://cran.r-project.org/web/packages/pinktoe/index.html
[http_cran.r-project.org_web_packages_randomForest_index.html]: http://cran.r-project.org/web/packages/randomForest/index.html
[http_cran.r-project.org_web_packages_ipred_index.html]: http://cran.r-project.org/web/packages/ipred/index.html
[http_cran.r-project.org_web_packages_varSelRF_index.html]: http://cran.r-project.org/web/packages/varSelRF/index.html
[http_cran.r-project.org_web_packages_lasso2_index.html]: http://cran.r-project.org/web/packages/lasso2/index.html
[http_cran.r-project.org_web_packages_lars_index.html]: http://cran.r-project.org/web/packages/lars/index.html
[http_cran.r-project.org_web_packages_elasticnet_index.html]: http://cran.r-project.org/web/packages/elasticnet/index.html
[http_cran.r-project.org_web_packages_glmpath_index.html]: http://cran.r-project.org/web/packages/glmpath/index.html
[http_cran.r-project.org_web_packages_penalized_index.html]: http://cran.r-project.org/web/packages/penalized/index.html
[http_cran.r-project.org_web_packages_pamr_index.html]: http://cran.r-project.org/web/packages/pamr/index.html
[http_cran.r-project.org_web_packages_earth_index.html]: http://cran.r-project.org/web/packages/earth/index.html
[http_cran.r-project.org_web_packages_gbm_index.html]: http://cran.r-project.org/web/packages/gbm/index.html
[http_cran.r-project.org_web_packages_boost_index.html]: http://cran.r-project.org/web/packages/boost/index.html
[http_cran.r-project.org_web_packages_GAMMoost_index.html]: http://cran.r-project.org/web/packages/GAMMoost/index.html
[http_cran.r-project.org_web_packages_mboost_index.html]: http://cran.r-project.org/web/packages/mboost/index.html
[http_cran.r-project.org_web_packages_e1071_index.html]: http://cran.r-project.org/web/packages/e1071/index.html
[http_cran.r-project.org_web_packages_kernlab_index.html]: http://cran.r-project.org/web/packages/kernlab/index.html
[http_cran.r-project.org_web_packages_klaR_index.html]: http://cran.r-project.org/web/packages/klaR/index.html
[http_cran.r-project.org_web_packages_BayesTree_index.html]: http://cran.r-project.org/web/packages/BayesTree/index.html
[http_www-stat.wharton.upenn.edu_edgeorge_Research_papers_BART_206_06.pdf]: http://www-stat.wharton.upenn.edu/~edgeorge/Research_papers/BART%206--06.pdf
[http_cran.r-project.org_web_packages_tgp_index.html]: http://cran.r-project.org/web/packages/tgp/index.html
[http_cran.r-project.org_web_packages_gafit_index.html]: http://cran.r-project.org/web/packages/gafit/index.html
[http_cran.r-project.org_web_packages_rgenoud_index.html]: http://cran.r-project.org/web/packages/rgenoud/index.html
[http_cran.r-project.org_web_packages_arules_index.html]: http://cran.r-project.org/web/packages/arules/index.html
[http_cran.r-project.org_web_packages_svmpath_index.html]: http://cran.r-project.org/web/packages/svmpath/index.html
[http_cran.r-project.org_web_packages_ROCR_index.html]: http://cran.r-project.org/web/packages/ROCR/index.html
[http_cran.r-project.org_web_packages_caret_index.html]: http://cran.r-project.org/web/packages/caret/index.html
[http_cran.r-project.org_web_packages_caretLSF_index.html]: http://cran.r-project.org/web/packages/caretLSF/index.html
[http_cran.r-project.org_web_packages_caretNWS_index.html]: http://cran.r-project.org/web/packages/caretNWS/index.html
[http_www-stat.stanford.edu_tibs_ElemStatLearn]: http://www-stat.stanford.edu/~tibs/ElemStatLearn/
[http_cran.r-project.org_web_packages_ElemStatLearn_index.html]: http://cran.r-project.org/web/packages/ElemStatLearn/index.html