Skip to content

Commit

Permalink
分类变量也可以计算相关性!
Browse files Browse the repository at this point in the history
  • Loading branch information
ixxmu committed Nov 19, 2024
1 parent 826633a commit 659cead
Showing 1 changed file with 15 additions and 0 deletions.
15 changes: 15 additions & 0 deletions docs/2024-11/分类变量也可以计算相关性_.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,15 @@
---
title: "分类变量也可以计算相关性!"
date: 2024-11-19T00:04:37Z
draft: ["false"]
tags: [
"fetched",
"R语言生信医学统计与科研"
]
categories: ["Acdemic"]
---
分类变量也可以计算相关性! by R语言生信医学统计与科研
------
<div><p>对于分类变量来说,如果是有序分类变量,可以使用Wilcoxon秩和检验,推断各等级强度的差异;对于无序分类变量,可以使用卡方检验/Fisher精确检验,比较频数分布的差异。上述这些统计检验,可以解释为列联表中行列变量是否存在统计学的差异(是否显著相互影响),但是无法给出相关性(关联)的强度或方向。</p><p>接下来,我们来讲解一下常见的用于统计列联表中行列变量相关性的系数:(1)Phi-coefficient,只适用于2X2的四格列联表,取值为[-1, 1]。其中,1表示强正相关,-1表示强负相关;(2)Pearson contingency coefficient,是Phi-coefficinet的矫正和推广,可以用于多维列联表资料,取值为[0,1];(3)Cramer`s V coefficient,是对行列数量不同时,对Pearson contingency coefficient的矫正,<span>取值为[0,1</span><span>],该系数不受样本量的限制。</span><br></p><p><span>在R中,可以通过 vcd 包来实现。<br></span></p><section><ul><li><li><li><li><li><li><li><li><li><li><li><li><li><li><li></ul><pre data-lang="properties"><code><span><span>&gt;</span> <span>library(vcd)</span></span></code><code><span><span>&gt;</span> <span>tab &lt;- table(lung$sex, lung$ph.ecog)</span></span></code><code><span><span>&gt;</span> <span>tab</span></span></code><code><span> </span></code><code><span> <span>0</span> <span>1 2 3</span></span></code><code><span> <span>1</span> <span>36 71 29 1</span></span></code><code><span> <span>2</span> <span>27 42 21 0</span></span></code><code><span><span>&gt;</span> <span>assocstats(tab)</span></span></code><code><span> <span>X^2</span> <span>df P(&gt; X^2)</span></span></code><code><span><span>Likelihood</span> <span>Ratio 1.6863 3 0.63998</span></span></code><code><span><span>Pearson</span> <span>1.3341 3 0.72105</span></span></code><code><span><br></span></code><code><span><span>Phi-Coefficient</span> : <span>NA </span></span></code><code><span><span>Contingency</span> <span>Coeff.: 0.076 </span></span></code><code><span><span>Cramer's</span> <span>V : 0.077</span></span></code></pre></section><p>解读:</p><ul><li><p><span>Likelihood Ratio:似然比卡方检验</span></p></li><li><p><span>Pearson:皮尔森卡方检验<br></span></p></li><li><p><span>Phi-Coefficient:只适用于2x2的四格列联表,这里不适用,输出NA</span></p></li><li><p><span>Contingency Coeff.:列联系数</span></p></li><li><p><span>Cramer's V:克莱姆相关系数</span></p></li></ul><p><mp-style-type data-value="3"></mp-style-type></p></div>
<hr>
<a href="https://mp.weixin.qq.com/s/jbM1MO2LWmoxZCTEUPmvFQ",target="_blank" rel="noopener noreferrer">原文链接</a>

0 comments on commit 659cead

Please sign in to comment.