How do I calculate the computational budget for a tuning run?

The skill provides a formula (number of trials × average training time × cross-validation folds) to help estimate total time and resources before starting an experiment.

Does this skill handle time-series cross-validation?

Yes, it specifically addresses time-series forecasting by mandating TimeSeriesSplit and providing strategies to avoid random K-fold CV which causes data leakage.

When should I choose Bayesian optimization over Random Search?

Bayesian optimization is recommended when you have more than 500 total fits or when model evaluation is computationally expensive, as it learns from previous trials to navigate the search space.

How does this skill assist with Optuna integration?

The skill provides code scaffolding for Optuna studies, including objective functions, trial suggestion patterns for various distributions, and pruning configurations for early stopping.

Can I use this for scikit-learn models?

Absolutely. It provides a decision tree for when to use native scikit-learn search classes versus when to wrap them in more advanced optimization libraries like Optuna.

Hyperparameter Tuning & Optimization

Name: Hyperparameter Tuning & Optimization
Author: andikarachman

byandikarachman

•

Data Science & ML

Optimizes machine learning model performance through structured hyperparameter tuning strategies and automated Bayesian search patterns.

This skill provides a comprehensive workflow reference for hyperparameter optimization within machine learning experiments. It guides developers through critical decision points including strategy selection (Grid, Random, Bayesian, or Successive Halving), search space design for popular frameworks like XGBoost and LightGBM, and budget estimation. By integrating Optuna-based Bayesian optimization and convergence diagnostics, it transforms the tuning process from manual trial-and-error into a systematic, reproducible engineering task, ensuring models reach peak performance without wasted computational resources.

Key Features

01Ready-to-use Optuna boilerplate for automated Bayesian optimization and trial pruning

029 GitHub stars

03Paradigm-specific tuning guidance for time-series, anomaly detection, and temporal data

04Advanced result analysis patterns for convergence and parameter importance diagnostics

05Decision tree for selecting optimal tuning strategies based on computational budget

06Pre-defined search space templates for supervised and unsupervised models

Use Cases

01Designing robust hyperparameter search spaces for high-dimensional gradient boosting models

02Transitioning from basic GridSearch to efficient Bayesian optimization using Optuna

03Managing computational costs through successive halving and budget-constrained tuning

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add andikarachman/data-science-plugin tuning-hyperparameters

For use in Claude.ai and ChatGPT

Key Features

01Ready-to-use Optuna boilerplate for automated Bayesian optimization and trial pruning

029 GitHub stars

03Paradigm-specific tuning guidance for time-series, anomaly detection, and temporal data

04Advanced result analysis patterns for convergence and parameter importance diagnostics

05Decision tree for selecting optimal tuning strategies based on computational budget

06Pre-defined search space templates for supervised and unsupervised models

Use Cases

01Designing robust hyperparameter search spaces for high-dimensional gradient boosting models

02Transitioning from basic GridSearch to efficient Bayesian optimization using Optuna

03Managing computational costs through successive halving and budget-constrained tuning