home / skills / dkyazzentwatwa / chatgpt-skills / classification-helper

classification-helper skill

safe

This skill helps you train and evaluate classification models with automatic model selection, hyperparameter tuning, and comprehensive metrics.

npx playbooks add skill dkyazzentwatwa/chatgpt-skills --skill classification-helper

Review the files below or copy the command above to add this skill to your agents.

Files (3)

SKILL.md

869 B

---
name: classification-helper
description: Quick classifier training with automatic model selection, hyperparameter tuning, and comprehensive evaluation metrics.
---

# Classification Helper

Train and evaluate classification models with automatic model selection.

## Features

- **Auto Model Selection**: Compare multiple classifiers
- **Hyperparameter Tuning**: Grid/random search
- **Evaluation Metrics**: Accuracy, precision, recall, F1, ROC-AUC
- **Cross-Validation**: K-fold validation
- **Confusion Matrix**: Detailed error analysis
- **Feature Importance**: Top predictive features
- **Model Export**: Save trained models

## CLI Usage

```bash
python classification_helper.py --data train.csv --target class --test test.csv --output model.pkl
```

## Dependencies

- scikit-learn>=1.3.0
- pandas>=2.0.0
- numpy>=1.24.0
- matplotlib>=3.7.0
- seaborn>=0.12.0

Overview

This skill provides a streamlined workflow to train, tune, and evaluate classification models with minimal setup. It automatically compares multiple candidate algorithms, runs hyperparameter searches, and produces comprehensive evaluation reports. The tool also supports k-fold cross-validation, confusion matrices, feature importance summaries, and model export for deployment.

How this skill works

You supply a labeled dataset and target column; the skill runs predefined candidate classifiers and ranks them by performance. It performs grid or random hyperparameter tuning and k-fold cross-validation to reduce overfitting. The output includes accuracy, precision, recall, F1, ROC-AUC, a confusion matrix, top predictive features, and a serialized model file for reuse.

When to use it

You need a fast, reproducible baseline for a new classification problem.
You want automated model selection without hand-crafting many experiments.
You need consistent cross-validated metrics for reporting or model comparison.
You want feature importance and error analysis to inform data decisions.
You need to export a trained model for production or further testing.

Best practices

Prepare and clean data in advance: handle missing values, encode categoricals, and scale features as needed.
Set a sensible search space for hyperparameters to balance compute cost and performance.
Use stratified k-fold when classes are imbalanced to get reliable validation scores.
Inspect confusion matrices and per-class metrics, not just overall accuracy.
Save preprocessing pipeline with the model to ensure consistent inference in production.

Example use cases

Quickly compare logistic regression, random forest, and gradient boosting on a new binary classification task.
Tune hyperparameters of candidate models using grid or random search with k-fold validation for robust selection.
Generate a validation report with ROC-AUC, F1, and confusion matrix to support model review meetings.
Identify top predictive features to guide feature engineering or reduce dimensionality.
Export the final pipeline and model artifact (e.g., model.pkl) for deployment in an inference service.

FAQ

What input formats are supported?

Use CSV files or pandas DataFrames; specify the target column and optional test split files.

Can I add custom classifiers or preprocessors?

Yes. Plug in additional scikit-learn compatible estimators or preprocessing steps into the workflow before running selection.

How does it handle imbalanced classes?

Use stratified k-fold validation and class-weighted estimators or sampling techniques. The tool reports per-class metrics to help evaluate imbalance.