TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Understanding Cohen's Kappa in Machine Learning

6 点作者 CarrieLab超过 3 年前

1 comment

CarrieLab超过 3 年前
I often see subtle misuses of interrater reliability metrics.<p>For example, imagine you&#x27;re running a Search Relevance task, where search raters label query&#x2F;result pairs on a 5-point scale: Very Relevant (+2), Slightly Relevant (+1), Okay (0), Slightly Irrelevant (-1), Very Irrelevant (-2).<p>Marking &quot;Very Relevant&quot; vs. &quot;Slightly Relevant&quot; isn&#x27;t a big difference, but &quot;Very Relevant&quot; vs. &quot;Very Irrelevant&quot; is. However, most IRR calculations don&#x27;t take this kind of ordering into account, so it gets ignored!<p>Cohen&#x27;s kappa is a rather simplistic and flawed metric, but a good starting point to understanding interrater reliability metrics.