NIR Spectroscopy Prediction API
May 2025The LabIn progress
A PLS regression model predicting moisture content from near-infrared spectra, served as an API.
- Trained a PLS regression model predicting moisture content from near-infrared spectra across 201 wavelengths per scan.
- Grouped repeat scans of the same sample so no sample leaked between the training and test sets.
- Containerised with Docker and deployed as a REST API on AWS Lambda.
- Automated build and deployment through GitHub Actions.
- Python
- scikit-learn
- Docker
- AWS Lambda
- GitHub Actions