Native Emergent Manifold Interrogation (NEMI) is a method for finding regions of interest in large, noisy, nonlinear data, especially in the Earth sciences. It combines manifold learning, clustering, and ensembling to highlight structure without relying on narrow statistical assumptions.
The docs follow the Diataxis idea: separate learning, tasks, facts, and background so you can open the right page for what you need now.
| You want to... | Start here |
|---|---|
| Do something specific (install, build docs) | Installation |
| Learn by doing a minimal, end-to-end run | Tutorial: first analysis |
| Understand the method and workflow | About NEMI |
| Look up classes, functions, and parameters | API reference |
NEMI generalises the approach in Sonnewald et al. (2020) (plankton ecosystems) to larger datasets and arbitrary data sources, with a hierarchical pipeline (fewer tuning knobs, useful from global to regional scales). It drops fixed field-specific benchmarks in favour of a field-agnostic option and supports several ways to quantify uncertainty in the final cluster assignment. For the full conceptual picture, see About NEMI.