nirs4all: from data to deployed predictions

The common user and developer guide follows the same workflow in Python, R, JavaScript/WASM, MATLAB/Octave, Rust and the CLI: load data, compare recipes, select and refit a model, save fitted state, and predict.

Start with a task and choose a supported language and runtime profile. The interface matrix identifies each product’s capabilities and the qualified directions for transferring fitted models.

Common user and developer guide

Chapters 0–13: installation, data, pipelines, evaluation, results, deployment, methods, language interfaces and contribution.

Learn to build, understand and reuse pipelines

Complete tutorials

Run the same documented workflow in Python, R, Node/WASM, MATLAB/Octave and the CLI.

11. Progress from a baseline to an advanced workflow

Interfaces and portability

Choose the Python SDK or portable Core, then check the supported data, artifact and runtime profile.

0. Choose the tools for your language

Legacy — Python

Previous Python guides, concepts, examples, developer documentation and detailed API references.

Legacy — Python documentation

Start with your next task

Task	Guide
Install and run a first pipeline	1. Run your first complete experiment
Load spectra, metadata or multiple modalities	4. Describe your measurements and align your sources
Compare pipelines and evaluate predictions	5. Build a pipeline you can explain and 6. Measure whether your model answers the question
Inspect results and saved workspaces	7. Read the result and keep a useful experiment record
Save fitted state and predict in another process or language	8. Save a predictor and use it on new observations
Choose a language interface	10. Run the same recipe in your language
Contribute a controller, reader or binding	12. Add an operation or contribute an example
Check exact API names and options	13. Find a node, parameter, example or solution

Documentation