Scholarlink

It’s great to do machine learning, building models, getting predictions, and finding interesting associations. But how do we know if what we’ve done is actually right? This is where performance metrics come in. There are a variety of ways of measuring characteristics of ML models that can help us understand whether they’re doing what we want and if they’re right. If our results aren’t actually meaningful, we don’t want to use them. Sometimes it’s tempting to just take the results at face value and assume they’re good. See the sidebar for more on this bad habit.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

How’d We Do? Measuring the Performance of ML Techniques

  • 章节

  • 发表日期：

    2025年

查看全文
  • Kelly P. Vincent

摘要

It’s great to do machine learning, building models, getting predictions, and finding interesting associations. But how do we know if what we’ve done is actually right? This is where performance metrics come in. There are a variety of ways of measuring characteristics of ML models that can help us understand whether they’re doing what we want and if they’re right. If our results aren’t actually meaningful, we don’t want to use them. Sometimes it’s tempting to just take the results at face value and assume they’re good. See the sidebar for more on this bad habit.