make-mlops-easy 0.1.0

Make MLOps easier with an AI-powered framework

Easy MLOps

Make MLOps easier with an AI-powered framework that abstracts all phases of an MLOps pipeline.

Overview

Easy MLOps is a comprehensive framework that simplifies the entire machine learning operations pipeline. With a single command, you can:

• Data Preprocessing: Automatic handling of missing values, feature scaling, and categorical encoding
• Model Training: Intelligent model selection and hyperparameter tuning
• Model Deployment: Automated model packaging and endpoint creation
• Model Observability: Built-in monitoring, logging, and performance tracking

Installation

From Source

git clone https://github.com/umbertocicciaa/easy-mlops.git
cd easy-mlops
pip install -e .

Using pip (once published)

pip install easy-mlops

Using Docker

Build the image locally and expose the CLI:

docker build -t easy-mlops .
docker run --rm easy-mlops --help

Mount data and persist model artifacts by binding host directories:

docker run --rm \
  -v "$(pwd)/data:/data" \
  -v "$(pwd)/models:/app/models" \
  easy-mlops train /data/dataset.csv --target price

The container entrypoint is easy-mlops, so any CLI subcommand can be passed as usual.

Quick Start

1. Train a Model

Train a machine learning model with a single command:

easy-mlops train data.csv --target price

This will:

• Load and preprocess your data
• Automatically detect the problem type (classification/regression)
• Train an appropriate model
• Deploy the model with all artifacts
• Set up monitoring and logging

2. Make Predictions

Use your deployed model to make predictions:

easy-mlops predict new_data.csv models/deployment_20240101_120000

3. Check Model Status

Get information about your deployed model:

easy-mlops status models/deployment_20240101_120000

4. View Observability Reports

Generate detailed monitoring reports:

easy-mlops observe models/deployment_20240101_120000

CLI Commands

easy-mlops train

Train a machine learning model on your data.

Usage:

easy-mlops train DATA_PATH [OPTIONS]

Options:

• -t, --target TEXT: Name of the target column
• -c, --config PATH: Path to configuration file (YAML)
• --no-deploy: Skip deployment step (only train the model)

Examples:

# Basic training
easy-mlops train data.csv --target label

# With custom configuration
easy-mlops train data.json -t price -c config.yaml

# Train without deploying
easy-mlops train data.csv --target label --no-deploy

easy-mlops predict

Make predictions using a deployed model.

Usage:

easy-mlops predict DATA_PATH MODEL_DIR [OPTIONS]

Options:

• -c, --config PATH: Path to configuration file (YAML)
• -o, --output PATH: Output file for predictions (JSON format)

Examples:

# Basic prediction
easy-mlops predict new_data.csv models/deployment_20240101_120000

# Save predictions to file
easy-mlops predict data.json models/latest -o predictions.json

easy-mlops status

Get status and metrics for a deployed model.

Usage:

easy-mlops status MODEL_DIR [OPTIONS]

Options:

• -c, --config PATH: Path to configuration file (YAML)

Example:

easy-mlops status models/deployment_20240101_120000

easy-mlops observe

Generate observability report for a deployed model.

Usage:

easy-mlops observe MODEL_DIR [OPTIONS]

Options:

• -c, --config PATH: Path to configuration file (YAML)
• -o, --output PATH: Output file for the report

Examples:

# View report in terminal
easy-mlops observe models/deployment_20240101_120000

# Save report to file
easy-mlops observe models/latest -o report.txt

easy-mlops init

Initialize a new MLOps project with default configuration.

Usage:

easy-mlops init [OPTIONS]

Options:

• -o, --output PATH: Output path for the configuration file (default: mlops-config.yaml)

Example:

easy-mlops init -o my-config.yaml

Configuration

Easy MLOps uses YAML configuration files to customize behavior. Generate a default configuration:

easy-mlops init

Configuration Structure

preprocessing:
  handle_missing: drop  # Options: drop, mean, median, mode
  scale_features: true
  encode_categorical: true

training:
  test_size: 0.2
  random_state: 42
  cv_folds: 5
  model_type: auto  # Options: auto, random_forest_classifier, random_forest_regressor, logistic_regression, linear_regression

deployment:
  output_dir: ./models
  save_format: joblib
  create_endpoint: false

observability:
  track_metrics: true
  log_predictions: true
  alert_threshold: 0.8

Features

Automatic Data Preprocessing

• Missing Value Handling: Multiple strategies (drop, mean, median, mode)
• Feature Scaling: StandardScaler for numerical features
• Categorical Encoding: Automatic label encoding for categorical variables
• Multiple Format Support: CSV, JSON, and Parquet files

Intelligent Model Training

• Problem Type Detection: Automatically detects classification vs regression
• Model Selection: Chooses appropriate algorithms based on your data
• Cross-Validation: Built-in k-fold cross-validation
• Performance Metrics: Comprehensive metrics for model evaluation

Streamlined Deployment

• Model Artifacts: Saves model, preprocessor, and metadata
• Versioning: Timestamped deployments for easy tracking
• Endpoint Creation: Optional prediction endpoint script generation
• Reproducibility: All components saved for consistent predictions

Built-in Observability

• Metrics Tracking: Automatic logging of model performance metrics
• Prediction Logging: Track all predictions for analysis
• Performance Monitoring: Detect metric degradation over time
• Reporting: Generate detailed monitoring reports

Supported Data Formats

• CSV (.csv)
• JSON (.json)
• Parquet (.parquet)

Supported Model Types

Classification

• Random Forest Classifier
• Logistic Regression

Regression

• Random Forest Regressor
• Linear Regression

Project Structure

easy-mlops/
├── easy_mlops/
│   ├── __init__.py
│   ├── cli.py                 # CLI interface
│   ├── pipeline.py            # Main pipeline orchestrator
│   ├── config/                # Configuration management
│   ├── preprocessing/         # Data preprocessing
│   ├── training/              # Model training
│   ├── deployment/            # Model deployment
│   └── observability/         # Model monitoring
├── pyproject.toml
├── requirements.txt
└── README.md

Requirements

• Python >= 3.8
• click >= 8.0.0
• pandas >= 1.3.0
• scikit-learn >= 1.0.0
• joblib >= 1.0.0
• pyyaml >= 6.0
• rich >= 10.0.0

Development

Running Tests

pytest

Code Formatting

black easy_mlops/

Documentation

Project documentation is powered by MkDocs and the Material theme. Source files live in docs/, and the site configuration is mkdocs.yml.

# Serve docs locally (http://localhost:8000)
make docs-serve

# Build static site into site/
make docs-build

The GitHub Actions workflow .github/workflows/docs.yml automatically builds and deploys the site to GitHub Pages on pushes to main. Enable GitHub Pages in the repository settings (source: “GitHub Actions”). For manual publishing or previews outside CI, run:

make docs-deploy

Type Checking

mypy easy_mlops/

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Author

Umberto Ciccia

Links

• GitHub: https://github.com/umbertocicciaa/easy-mlops
• Issues: https://github.com/umbertocicciaa/easy-mlops/issues