LDA vs. QDA vs. Logistic Regression

1. Classification Model比较

. LDA QDA Logistic KNN
Parameters Less More Less Non-parametric
Decision. Linear Quatratic Linear Non-linear
Bias High Low High Low

对于两个Linear Decision Boundary的方法而言(LDA & Logistic regression),当X满足Guassian Distribution且standardized deviation相等时,使用LDA有更小的偏差。否则使用Logistic regression有更小的偏差。

2. 读书笔记

以下笔记摘录自An Introduction to Statistical Learning by James, Witten, Hastie and Tibshirani Statistical Learning Cheat Sheet

相关