基于平衡准确率和规模的决策树剪枝算法

决策树剪枝是决策树分类学习中的重要步骤,可降低决策树复杂程度和提高决策树泛化能力,从而提高决策树识别精度和效率。通过利用系数函数综合决策树的错误率和规模,形成决策树剪枝标准,在系数函数的参数合适选取,采用自底向上遍历过程逐一进行判断剪枝。实验结果表明,综合考虑决策树的分类预测准确率和决策树的规模大小,BASP剪枝算法能够获得更好的剪枝效果。

Pruning Algorithm of Decision Tree by Balance of Accuracy and Size

Pruning is an important step of decision tree learning, which can reduce the complexity of decision tree and improve its generalization ability to gain the accuracy effectively and efficiently. Definition of function, which combines the error rate and size of decision tree, serves as a criterion for decision tree pruning. After proper selection of coefficient of the function, the procedure of bottom-up traverse is adopted for the decision tree to prune by the criterion, resulting in good accuracy and performance.