Matminer property regression starter
A framework for discovering, compiling, and validating reusable skills for scientific agents.
npx -y skills add ma-compbio-lab/SkillFoundry --skill matminer-property-regression-starterAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
SKILL.md
1.0 KB, as published. Nobody here has run it
Matminer Property Regression Starter
Use this skill to featurize simple formulas with Matminer and fit a deterministic toy regression model for materials-property prediction.
What This Skill Does
- converts formulas into composition features with Matminer
- fits a small deterministic regression model
- reports feature count, train fit quality, and holdout predictions
When To Use It
- when a user asks for a starter on materials-property prediction
- when you need a lightweight local template before connecting to a larger materials dataset
- when you want a runnable example for the
materials-property-predictiontaxonomy leaf
Run
./slurm/envs/materials/bin/python skills/materials-science-and-engineering/matminer-property-regression-starter/scripts/run_matminer_property_regression.py --out scratch/materials/property_regression_summary.json
Notes
- This is a toy local regression starter, not a benchmark-quality predictive model.
- Pair it with larger curated datasets later if you need scientific performance claims.