给出对决策树与随机森林的认识。主要分析决策树的学习算法：信息增益和ID3、C4.5、CART树，然后给出随机森林。
决策树中，最重要的问题有3个:
1. 特征选择。即选择哪个特征作为某个节点的分类特征；
2. 特征值的选择。即选择好特征后怎么划分子树；
3. 决策树出现过拟合怎么办？
下面分别就以上问题对决策树给出解释。决策树往往是递归的选择最优特征，并根据该特征对训练数据进行分割。(The understanding of decision tree and random forest is given. This paper mainly analyzes the learning algorithm of decision tree: information gain and ID3, C4.5, CART tree, and then give the random forest.
Among the decision trees, there are 3 of the most important issues.
1. feature selection. Which is to choose which feature as the classification of a node;
2. the selection of eigenvalues. That is, how to divide the subtrees after the selection of the good features.
3. how to do the fitting of the decision tree?
The following questions are explained on the decision tree respectively. The decision tree is often the optimal feature of the recursive selection, and the training data are segmented according to the feature.)