east-py-datascience

Run optimization and ML workflows with Scikit-learn, XGBoost, and SHAP on East.

1|Updated Nov 2, 2025

Search Tags:#optimization #preprocessing #data-science #machine-learning #explainability #east #SHAP

Ready to run

npx skills add https://github.com/elaraai/east-plugin --skill east-py-datascience

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: east-py-datascience
Source: https://github.com/elaraai/east-plugin/tree/main/skills/east-py-datascience
Command: npx skills add https://github.com/elaraai/east-plugin --skill east-py-datascience

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The East Data Science Skill provides a unified toolkit for data science within East, enabling optimization, ML modeling, preprocessing, and explainability using a single interface.

Core Features & Use Cases

MADS, Optuna, SimAnneal, ALNS for derivative-free, Bayesian, simulated annealing, and large neighborhood search optimization.
Sklearn, Scipy, XGBoost, LightGBM, NGBoost, Torch, Lightning, GP, MAPIE, and Shap modules for ML, statistics, probabilistic modeling, and model explainability.
Use cases include building objective functions in East, running optimizers, training models, applying conformal predictions, and generating SHAP explanations across supported models.

Quick Start

Define a tiny optimization problem with MADS or Optuna, run the optimizer, then train a small model with the East Py-DS bridge and generate SHAP explanations.

Frequently Asked Questions about east-py-datascience

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run optimization with derivative-free methods in East?▼

Derivative-free optimization in East uses MADS, Optuna, simulated annealing, or ALNS to explore parameter spaces without gradients. Define an objective function, select an optimizer, and execute it through the East Py-DS bridge to find optimal values for your problem.

Can I train machine learning models with XGBoost, LightGBM, and PyTorch in East?▼

Yes. East Py-DS supports XGBoost, LightGBM, NGBoost, PyTorch, and Lightning for model training. Use the unified interface to build, train, and evaluate models across these frameworks without switching tools or managing separate installations.

How do I explain model predictions with SHAP in East?▼

SHAP explanations generate feature importance and prediction breakdowns for supported models trained in East. Apply SHAP directly to your fitted model to produce interpretable, model-agnostic explanations of individual predictions and overall feature contributions.

Does East support conformal predictions for uncertainty quantification?▼

Yes. East Py-DS includes MAPIE for conformal prediction, enabling you to attach calibrated confidence intervals to model outputs. This provides distribution-free uncertainty estimates alongside point predictions across regression and classification tasks.

What preprocessing and statistical tools are available in East for data science?▼

Preprocessing and statistical capabilities span Scikit-learn and Scipy within East Py-DS, covering scaling, transformation, hypothesis testing, and statistical modeling. These tools integrate seamlessly with optimization and ML components in a single workflow.

Do I need additional Python packages beyond East to use data science features?▼

No. East Py-DS bundles all required libraries—Scikit-learn, XGBoost, LightGBM, PyTorch, SHAP, MAPIE, and others—through the East platform and @elaraai/east-py-datascience bridge. No separate installations are needed.