skills.rest
MonumentalSystems
MonumentalSystemsOFFICIAL
Explore all skills by this author →
WebsiteGitHub

ml-best-practices

Guide model selection, feature engineering, and hyperparameter tuning for ML workflows.

17|5|Updated Feb 6, 2026
Search Tags:#machine learning#feature engineering#hyperparameter tuning#model interpretability#model selection#cross-validation#evaluation metrics
One-click install
npx skills add https://github.com/MonumentalSystems/Atlas-Agent-Teams --skill ml-best-practices-monumentalsystems
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ml-best-practices
Source: https://github.com/MonumentalSystems/Atlas-Agent-Teams/tree/main/teams/data-science/skills/ml-best-practices
Command: npx skills add https://github.com/MonumentalSystems/Atlas-Agent-Teams --skill ml-best-practices-monumentalsystems

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill resolves inconsistent ML outcomes by guiding practitioners through disciplined model selection, feature engineering, hyperparameter tuning, and evaluation metric choice.

Core Features & Use Cases

  • Model Selection Guidelines: Matches supervised, unsupervised, and reinforcement learning approaches to problem type and constraints such as interpretability, latency, and training time.
  • Feature Engineering Techniques: Covers scaling, encoding, text vectorization, embeddings, and feature selection methods to improve signal quality before training.
  • Hyperparameter Tuning Strategies: Provides practical search strategies (grid/random/bayesian/evolutionary) and tuning best practices like cross-validation and early stopping.
  • Evaluation Metrics and Validation Methods: Chooses appropriate regression/classification metrics and validation schemes (including nested CV and time series splits) to reduce bias and leakage.
  • Model Interpretation Methods: Recommends explainability techniques such as SHAP, LIME, permutation importance, and partial dependence plots.

Quick Start

Use ml-best-practices to design an end-to-end plan for selecting a model, engineering features, tuning hyperparameters with cross-validation, and reporting metrics for a tabular classification problem with imbalanced labels.

Frequently Asked Questions about ml-best-practices

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

FAQPage Schema
How do I choose the right machine learning model for my classification problem?▼

For model selection, match supervised, unsupervised, or reinforcement learning approaches to your problem type and practical constraints like interpretability, latency, and training time. This ensures framework-appropriate modeling choices for your specific classification task.

What is the best way to tune hyperparameters and prevent overfitting?▼

The best way to tune hyperparameters is to apply search strategies like grid, random, bayesian, or evolutionary search. Combine these with cross-validation and early stopping to improve machine learning performance and robustness while preventing overfitting.

How does cross-validation reduce bias and data leakage in time series tasks?▼

Cross-validation reduces bias and leakage by applying appropriate validation schemes like nested CV and time series splits. These validation methods ensure that temporal data is evaluated correctly, preventing future information from influencing past predictions during model evaluation.

What feature engineering techniques improve signal quality for text data?▼

Feature engineering techniques that improve signal quality for text data include text vectorization and embeddings. Additionally, applying scaling, encoding, and feature selection methods before training enhances the overall signal quality across tabular and text data tasks.

Which model interpretability methods work best for explaining predictions?▼

Model interpretability methods that effectively explain predictions include SHAP, LIME, permutation importance, and partial dependence plots. These explainability techniques help practitioners understand model behavior and interpret results reliably across different data tasks.

What evaluation metrics should I use for imbalanced labels in tabular data?▼

Evaluation metrics for imbalanced labels in tabular data must be chosen based on the specific regression or classification problem. Selecting appropriate validation schemes alongside these metrics reduces bias and ensures reliable reporting for imbalanced classification tasks.

Related Skills

Explore More Data & Analytics Skills→
lucifertrjlucifertrj
17

nltk-nlp-text-cleaning

Clean, tokenize, and normalize English text with NLTK for NLP pipelines.

Community
Advanced
timurgaleevtimurgaleev
5

canary

Monitor post-deploy web apps for defects and performance regressions.

Community
Advanced
nferrer-devnferrer-dev

scikit-learn

Guide machine learning tasks with scikit-learn pipelines, preprocessing, and cross-validation.

Community
Advanced
GiacomoSaccaggiGiacomoSaccaggi
3

scomp-link

Automate machine learning workflows from data profiling to deployment.

Community
Advanced