Does this skill support Scikit-learn?

Yes, JAXSR provides estimators that are compatible with Scikit-learn's cross-validation, GridSearchCV, and Pipeline protocols.

What is JAXSR used for?

JAXSR is used to discover simple, interpretable mathematical formulas (algebraic expressions) that explain the relationship between input features and output data.

Can I use JAXSR for experimental design?

Absolutely. JAXSR includes a DOEStudy module for creating designs like Latin Hypercube and suggesting the most informative next points to sample.

How does JAXSR handle uncertainty quantification?

It offers multiple methods including standard confidence intervals, Bayesian Model Averaging (BMA), and distribution-free conformal prediction.

Can I enforce physical laws in the discovered models?

Yes, JAXSR allows you to add constraints such as monotonicity, convexity, and specific output bounds to ensure the model remains physically consistent.

JAXSR Symbolic Regression

Name: JAXSR Symbolic Regression
Author: jkitchin

byjkitchin

•

数据科学与机器学习

Discovers interpretable algebraic equations from data using JAX-powered sparse optimization and Design of Experiments (DOE) workflows.

JAXSR is a high-performance symbolic regression assistant designed to help researchers and engineers discover the simplest mathematical models that accurately explain their data. It leverages JAX for computational efficiency and follows ALAMO-style methodologies to build rich candidate bases and select optimal models through sparse optimization. The skill provides an interactive assistant mode that guides users through problem characterization, basis function selection, and the application of physical constraints like monotonicity or boundary conditions. Beyond model fitting, JAXSR supports rigorous uncertainty quantification and complete Design of Experiments (DOE) lifecycles, from initial design to automated reporting in Excel or Word.

主要功能

01Seamless integration with Scikit-learn pipelines and automated reporting in Excel/Word/LaTeX.

02Support for physical constraints including monotonicity, bounds, and sign constraints.

03Automated discovery of interpretable algebraic expressions from datasets.

0434 GitHub stars

05Comprehensive uncertainty quantification using BMA, conformal prediction, and bootstrapping.

06Integrated Design of Experiments (DOE) for active learning and next-point suggestions.

使用场景

01Creating interpretable response surface models for engineering optimization and DOE.

02Deriving chemical kinetics or physical rate laws from experimental measurements.

03Identifying key features and governing equations in complex multi-dimensional datasets.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jkitchin/jaxsr jaxsr

For use in Claude.ai and ChatGPT

Download Skill