决策树实践小技巧

1.当sample较少，features较多时，容易产生过拟合。所以进行降为预处理（PCA，ICA，feature selection）很重要。
2.可以用max_depth=3作为树的初始深度，export函数来可视化拟合的过程，对拟合过程有个初步体验，此后再增加树的深度。
3.用法max_depth来控制树的大小，防止过拟合。（Remember that the number of samples required to populate the tree doubles for each additional level the tree grows to. ）
4.回归时max_feature的值选择n_feature 较好，分类时max_feature的值选择sqrt(n_features)较好。Empirical good default values are max_features=n_features for regression problems, and max_features=sqrt(n_features) for classification tasks (where n_features is the number of features in the data).
5.一般来说，将max_depth设置为None，min_samples_split设置为2结果较好。（Good results are often achieved when setting max_depth=None in combination with min_samples_split=2）
6.最好进行交叉验证：The best parameter values should always be cross-validated.
7.The size of the coding is at most n_estimators * 2 ** max_depth, the maximum number of leaves in the forest.
https://scikit-learn.org/stable/modules/tree.html#tree
https://scikit-learn.org/stable/modules/ensemble.html#forest