machine-learning_skill

This skill helps you build, evaluate, and tune machine learning models using scikit-learn for classification, regression, and clustering.

Python

5

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill pluginagentmarketplace/custom-plugin-ai-data-scientist --skill machine-learning

SKILL.md4.9 KB

Overview

This skill provides practical tools and recipes for supervised and unsupervised machine learning using scikit-learn. It covers building, training, evaluating, and selecting models for classification, regression, and clustering. Use it to accelerate model development with ready patterns for pipelines, cross-validation, hyperparameter tuning, and feature engineering.

How this skill works

The skill supplies code patterns and guidance that walk through data splitting, model fitting, prediction, and evaluation using scikit-learn estimators. It includes examples for classification, regression, and clustering, plus utilities for scaling, encoding, pipelines, cross-validation, and GridSearch hyperparameter tuning. Results are validated with standard metrics and best-practice checks to avoid common pitfalls like data leakage and overfitting.

When to use it

Building a baseline or production-ready classifier or regressor quickly with scikit-learn
Comparing model families (linear, tree-based, boosting, SVM) to pick the right approach
Tuning hyperparameters and estimating generalization via cross-validation
Clustering data and selecting cluster count using the elbow method or density-based clustering
Creating reproducible pipelines that include preprocessing and modeling steps

Best practices

Split data into train/test before any preprocessing to prevent data leakage
Use cross-validation (k-fold) for reliable performance estimates and to tune hyperparameters
Scale features for distance-based methods (SVM, KMeans) and apply consistent transforms with Pipelines
Address class imbalance with class weights or resampling (SMOTE) and monitor class-wise metrics
Persist models and preprocessing objects (joblib/pickle) to reproduce production predictions

Example use cases

Train a RandomForest classifier with GridSearchCV and report weighted F1 and ROC-AUC
Fit a GradientBoostingRegressor, compute MAE and R², and compare against a linear baseline
Use KMeans with an elbow plot to choose k and assign cluster labels for customer segmentation
Build a Pipeline that scales features and trains a classifier to avoid manual transform errors
Apply cross_val_score to validate model stability across folds before deployment

FAQ

Start with a simple, interpretable baseline like logistic regression, then try tree-based models (Random Forest, XGBoost) for non-linear patterns.

How do I prevent data leakage?

Always split into train/test before preprocessing and incorporate transforms inside a Pipeline so fit/transform occurs only on training data during cross-validation.