该方法考虑了模型复杂度的降低，决策树生成学习局部模型，而决策树剪枝学习

This method takes into account the reduced complexity, decision trees to generate partial learning model, the overall model learning decision tree pruning, pruning is performed using a regularization estimation model selection maximized with a minimum loss function principle.

This method takes into account the reduction of model complexity, the decision tree generates the learning local model, and the decision tree pruning learns the overall model, and the cutting principle of the loss function is to make the model selection by using the regularization maximization estimation.

This method takes into account the reduction of model complexity, decision tree generates learning local model, while decision tree pruning learning overall model, using the principle of loss function minimum pruning is using regularization maximization estimation to select model.<br>