Supposed that you have applied Decision Tree algorithm to your training set and validation set and you have got the following scenarios by choosing different depth for your algorithm. Which depth you will pick for your final model?[table]
Depth | Training error | Validation error 3 | 30% | 35% 4 | 26% | 30% 5 | 22% | 30% 6 | 15% | 35%
[/table]

Question

BlackTom AI · Accepted Answer

We start by comparing the validation errors across the different depths, since the goal is to select a model that generalizes well to unseen data.
Option a (depth 3) shows a validation error of 35%, which is worse than the other depths and indicates poorer generalization.
Option b (depth 4) yields a validation error of 30%. This is the lowest validation error among the options (tied with depth 5, see next point), suggesting better generalization than depth 3.
Option c (depth 5) also has a validation error of 30%, matching depth 4 in terms of generalization performance, but depth 5 uses a more complex model (lower training error of 22% vs 26% at depth 4), which can risk overfitting without improvement in validation.
Option d (depth 6) shows an increase in validation error to 35%, despite the training error being the smallest (15%), which is a classic sign of overfitting and worse generalization.

Putting the pieces together, depths 4 and 5 both achieve the best (lowest) validation error of 30%. Between these two, depth 4 is simpler (higher bias but lower variance), and it achieves the same validation performance as depth 5. Therefore, depth 4 is the preferred choice for the final model to balance generalization with model simplicity.

查看解析

登录即可查看完整答案

类似问题

In a decision tree, variables used closer to the root are generally:

Decision Trees split data to minimize which metric in classification?

Which of the following numbers is closest to the best parameter setting for the Decision Tree algorithm on the income data?

The best attribute to use at the root of a decision tree is that with the lowest entropy. TRUE or FALSE?

更多留学生实用工具

考试浏览器助手

风格化写作助手

论文查重助手

文献引用助手

课堂转译助手

课堂笔记助手

Quiz搜索助手

学校历年真题

智能刷题助手

智能匹配练习

希望你的学习变得更简单