Stop Misusing t-SNE and UMAP for Visual Analytics

2025-06-10

AI Summary

This paper identifies a widespread misuse of t-SNE and UMAP in visual analytics—specifically, inferring inter-cluster distances from their low-dimensional projections despite well-documented geometric distortions. Method: Through bibliometric analysis of 114 papers and semi-structured interviews with domain experts, augmented by qualitative coding and causal attribution, the study systematically uncovers six latent drivers underlying this misuse. Contribution/Results: It reveals “discursive absence” as the core issue: a lack of consensus on the geometric semantic boundaries of dimensionality reduction methods. Building on this, the paper proposes a practitioner-oriented framework comprising seven actionable recommendations and outlines key directions for future research. This work bridges critical theoretical and practical gaps in the normative use of dimensionality reduction techniques within visualization research and practice.

Technology Category

• Data Mining & Knowledge Management: Data Visualization & Summarization
• Machine Learning: Dimensionality Reduction/Feature Selection
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Web Mining and Content Analysis: Web data visualization
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Misuses of t-SNE and UMAP in visual analytics have become increasingly common. For example, although t-SNE and UMAP projections often do not faithfully reflect true distances between clusters, practitioners frequently use them to investigate inter-cluster relationships. In this paper, we bring this issue to the surface and comprehensively investigate why such misuse occurs and how to prevent it. We conduct a literature review of 114 papers to verify the prevalence of the misuse and analyze the reasonings behind it. We then execute an interview study to uncover practitioners' implicit motivations for using these techniques -- rationales often undisclosed in the literature. Our findings indicate that misuse of t-SNE and UMAP primarily stems from limited discourse on their appropriate use in visual analytics. We conclude by proposing future directions and concrete action items to promote more reasonable use of DR.

Problem

• Addressing misuse of t-SNE and UMAP in visual analytics
• Investigating why practitioners misinterpret cluster distances
• Proposing solutions to prevent improper dimensionality reduction use

Innovation

• Literature review of 114 papers
• Interview study with practitioners
• Propose future directions for DR