教你怎么用真实标签评估聚类效果，选对指标很关键。
How to evaluate clustering with ground truth?
- 用集匹配思想衡量聚类结果与真实标签的相似性
- 中心点指数（CI）直观易懂，适合整体评估聚类质量
- 按需选择：点级评估用ACC，大小不均时用配对集指数（PSI）
当存在真实标签时，可使用外部指标评估聚类效果。本文重点回顾基于集合匹配的外部有效性指标。推荐使用中心点指数（CI），因其为直观的聚类级别度量，结果易于解释。若需更精细的点级度量，则有更多选择：配对集指数（PSI）提供归一化得分，不受聚类规模影响；若所有数据点权重应相等，则聚类准确率（ACC）或其他集合匹配指标更为合适。
原文摘要 · Abstract (English)
External indexes can be used for cluster evaluation when ground truth is available. We review the most common external validity indexes focusing on set-matching-based measures. We recommend centroid index (CI), because it is an intuitive cluster-level measure with an explainable result. If we need a more fine-tuned, point-level measure, there are more choices. Pair-set index (PSI) provides a normalized score which is not biased by cluster sizes. If all points should matter equally, then clustering accuracy (ACC) or any other set-matching measure is suitable.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。