Hands-On Machine Learning with Scikit-Learn & TensorFlow Exercise Q&A Chapter06

Leonardo Liu

于 2019-03-21 22:16:23 发布

阅读量372

点赞数

CC 4.0 BY-SA版权

分类专栏： Python 机器学习 Scikit-Learn Hands-On ML with sklearn & TensorFlow Exercise Q&A 文章标签： Machine Learning HandsOn Decision Tree

本文链接：https://blog.youkuaiyun.com/leowinbow/article/details/88724999

本文是一篇关于机器学习实战中决策树的问答集锦，涉及了决策树的深度、节点纯度、过拟合与欠拟合的解决策略、训练时间复杂度以及预排序对训练速度的影响。内容包括：决策树的深度与实例数量的关系，节点基尼不纯度与父节点的比较，如何应对过拟合和欠拟合，以及大规模数据训练时间的预测。最后提到了在moon数据集上训练和优化决策树以及构建森林的方法。

Q1. What is the approximate deph of a Decision Tree trained (without restrictions) on a training set with 1 million instances?

A1: The depth of a well-balanced binary tree containing m leaves is equal to $log_{2}(m)$ , so when there is 1 million instances the approximate depth is $log_{2}(10^{6})$ $\approx$ 20.