Personality Space Explorer visualizes 50k+ celebrity personalities in an interactive 3D map. Built with React, Vite, Three.js, and react-three-fiber, it lets you fly through clustered personality space, search by name, filter by MBTI and Enneagram, and inspect detailed profiles. A Python pipeline (Pandas, scikit-learn, UMAP, HDBSCAN) was used to reduce high‑dimensional traits into 3D coordinates and labels clusters via the Claude API, exported as Parquet file for fast loading.

Wispr Flow: Dictation That Works Everywhere — Stop typing. Start speaking. 4x faster.

Stop typing. Start speaking. 4x faster.

Hey Product Hunt! I’m a college student getting into machine learning, with a long‑time interest in psychology and personality. I found the MBTI Celebrity Personality Types dataset on Kaggle and decided to apply the unsupervised learning techniques I’d just read about in a textbook. This side project maps 50k+ celebrities into a 3D personality space using UMAP for dimensionality reduction and HDBSCAN for clustering, then renders it with React + Three.js so you can explore clusters, search by name, and filter by MBTI/Enneagram. Dataset: https://www.kaggle.com/datasets/... Special thanks to the dataset creator on Kaggle, the UMAP and HDBSCAN authors, and the Claude API for cluster labeling. Feel free to share ideas for new features or how to generalize this approach to other themes and datasets. I'm happy to answer any questions and thanks for checking it out :)

Wispr Flow: Dictation That Works Everywhere — Stop typing. Start speaking. 4x faster.

Stop typing. Start speaking. 4x faster.

Personality Space Explorer

Explore 50k celebrity personalities in a 3D space

Explore 50k celebrity personalities in a 3D space