Short description: A reproducible MLOps project that analyzes AI tool adoption using feature engineering, clustering, and a neural-network classifier. The repository demonstrates modular code, experiment tracking with MLflow, and data versioning with DVC.

• License: LICENSE
• Python: Compatible with Python 3.8+

Add CI / coverage badges here when available (GitHub Actions, Codecov, etc.).

• Python 3.8+ (recommended using a virtualenv or venv)
• pip
• DVC (optional but recommended for reproducibility)
• MLflow (for experiment tracking)

# create a venv and activate
python -m venv .venv
source .venv/bin/activate

# install requirements
pip install -r requirements.txt

# Pull data artifacts tracked by DVC (requires configured remote)
dvc pull

If you do not use DVC remote, ensure data/raw/ai_adoption_dataset.csv is present.

mlflow ui --port 5000
# open http://localhost:5000 to view runs and metrics

python src/pipeline/main.py

This runs the end-to-end pipeline that:

• loads and validates data,
• engineers features and creates clusters,
• prepares training data,
• trains and evaluates a Keras classifier,
• logs parameters/metrics/artifacts to MLflow,
• saves models under models/.

Open any notebook in notebooks/ (e.g., EDA.ipynb, experiments.ipynb) with Jupyter Lab/Notebook for interactive EDA and prototyping:

jupyter lab notebooks/

• Modular pipeline: clear separation of data loading, preprocessing, feature engineering, clustering, and model training.
• ML models: TensorFlow/Keras-based neural classifier and clustering for segmentation.
• MLOps tools: experiment tracking with MLflow, data/model versioning with DVC.
• Reproducibility: pipeline designed to be repeatable and auditable.
• Visualization: notebooks and plotting utilities for EDA and results.

.
├── data/
│   ├── raw/                # Raw datasets (e.g., ai_adoption_dataset.csv)
│   └── processed/          # Processed datasets after cleaning/feature engineering
├── log/
│   └── logging.log         # Pipeline and application logs
├── models/                 # Saved models (e.g., model.keras, scaler.joblib)
├── notebooks/
│   ├── EDA.ipynb           # Exploratory Data Analysis notebook
│   ├── experiments.ipynb   # Experiment tracking notebook
│   └── exploratory.ipynb   # Main exploratory notebook
├── reports/
│   └── report_1.txt        # Generated data analysis/model reports
├── src/
│   ├── constant/           # Project-wide constants and configuration
│   ├── data/               # Data loading and validation utilities
│   ├── feature_engineering/# Feature engineering and transformation scripts
│   ├── model/              # Deep learning and ML model definitions
│   ├── pipeline/           # Pipeline orchestration and main scripts
│   ├── preprocessing/      # Data preprocessing and encoding scripts
│   ├── utils/              # Helper functions (saving/loading, splitting, scaling)
│   └── visualization/      # Visualization scripts
├── .dvcignore              # DVC ignore file
├── .gitignore              # Git ignore file
├── LICENSE                 # License file
├── requirements.txt        # Python dependencies
├── setup.py                # Python package setup
└── README.md               # Project documentation

Refer to src/pipeline/pipeline.py for the main pipeline flow and src/model/ for model definitions.

• Run a local training experiment and view metrics in MLflow:

# (1) Optional: run mlflow ui in another terminal
mlflow ui
# (2) Run the pipeline
python src/pipeline/main.py
# (3) Open MLflow UI and inspect the latest run

• To reproduce results using DVC (if dvc pipeline files are present):

# fetch remote data
dvc pull
# reproduce pipeline (if configured)
dvc repro

Contributions are welcome. Please:

1. Open an issue to discuss larger changes.
2. Create a branch for your feature/fix.
3. Open a pull request with tests and a clear description.

Style guidelines:

• Keep functions small and well-tested
• Add docstrings to public functions and classes
• Follow PEP8 (use black / flake8 if available)

There are no automated tests yet. Contributions adding tests (pytest) are highly appreciated.

• If MLflow does not show runs, ensure you're running the UI in the repository root, and that runs are logged to the default mlruns/ folder.
• If data files are missing, verify data/raw/ai_adoption_dataset.csv or run dvc pull if you use DVC remotes.

This project is released under the terms in LICENSE.

If you have questions or want to collaborate, open an issue or contact the repository owner.