CSE30246 Study Guide - Final Guide: Overfitting, Credit Score
6 Jan 2019
School
Department
Course
Professor
Document Summary
Machine learning final study guide: decision trees usually overfit against training data, need to reduce the size of the tree in order to generalize better through the process of pruning, decision trees can be, 1. Pre-pruned: stops growing the tree earlier before it perfectly classifies the training set: 2. In pre-pruning, if the cost of adding another variable to the decision tree from the current node is above cp, then the tree building does not continue if > cp, then stop. In post-pruning, cp value that corresponds to the lowest cross- validation error is used as the threshold for pruning the tree: strengths of decision trees, 1. Handles numeric and nominal features well: 3. Useful for both large and small datasets: 6. Efficient and low cost model: weaknesses of decision trees, 1. Splits biased towards features with a large number of levels: 2. Reliance on axis-parallel splits is limiting: 4.
