MLGuard

MLGuard is a lightweight machine learning experiment management and data validation toolkit designed to help data scientists quickly validate datasets, detect potential data leakage, and benchmark multiple machine learning models.

It provides a simple interface to run experiments and compare models with minimal setup.

---

Why MLGuard?

When working with machine learning pipelines, developers often face problems such as:

• Undetected data leakage
• Repeated model experimentation scripts
• Difficulty comparing models quickly
• Poor experiment organization

MLGuard helps solve these problems by providing:

• Automated data leakage detection
• Automatic problem type detection
• Built-in experiment manager
• Simple model comparison tools

---

Features

Data Leakage Detection

Detects potential target leakage by analyzing correlations between features and the target variable.

Experiment Manager

Runs multiple machine learning models automatically and evaluates their performance.

Model Comparison

Compares model performance using appropriate evaluation metrics.

Automatic Problem Detection

Detects whether the task is:

• Regression
• Classification

---

Installation

Install from PyPI:

pip install mlguardlabs

Install locally for development:

pip install -e .

---

Quick Example

Regression Example

from sklearn.datasets import fetch_california_housing
from mlguard import ExperimentManager

data = fetch_california_housing(as_frame=True)

X = data.data
y = data.target

exp = ExperimentManager()

exp.fit(X, y)

print(exp.compare())

Example output:

Running leakage detection...
No leakage detected

Detected problem type: regression

[('rf', 0.50), ('ridge', 0.74)]

---

Classification Example

from sklearn.datasets import load_iris
from mlguard import ExperimentManager

data = load_iris(as_frame=True)

X = data.data
y = data.target

exp = ExperimentManager()

exp.fit(X, y)

print(exp.compare())

---

Data Leakage Detection Example

from mlguard import LeakageDetector

detector = LeakageDetector()

detector.fit(X, y)

print(detector.report())

This prints features that have unusually high correlation with the target.

---

How MLGuard Works

MLGuard runs a lightweight validation and experimentation pipeline:

Dataset
   ↓
Leakage Detection
   ↓
Problem Type Detection
   ↓
Model Experiments
   ↓
Model Evaluation
   ↓
Model Comparison

This allows users to quickly answer:

Which model works best for my dataset?

---

Project Structure

mlguard/
│
├── experiments
│   └── experiment_manager.py
│
├── inspection
│   └── leakage_detector.py
│
├── metrics
│   ├── regression_metrics.py
│   └── classification_metrics.py
│
├── utils
│   └── problem_type.py
│
├── examples
├── tests

---

Example Workflow

Typical machine learning workflow using MLGuard:

EDA
↓
Feature Engineering
↓
Feature Selection
↓
MLGuard Experiment Manager
↓
Best Model Selection
↓
Hyperparameter Tuning
↓
Final Model

MLGuard helps simplify the experimentation stage.

---

Roadmap

Future planned features include:

• Cross-validation experiment engine
• Dataset inspector
• Automatic preprocessing pipelines
• Feature validation tools
• Experiment tracking

---

Contributing

Contributions are welcome!

Steps:

1. Fork the repository
2. Create a feature branch
3. Submit a pull request

---

License

MIT License © 2026 Tarun M