Rock protolith prediction app

ML model trained to predict if a rock is igneous or sedimentary based on major element geochemistry.

This model is provided as a web app which can be found at: http://protolith-app.s3-website-ap-southeast-2.amazonaws.com/

See also the modified serverless version of this app: https://github.com/RADutchie/Rock-protolith-predictor-serverless

To run locally

• Setup a virtualenv using conda or venv running Python 3.6.10
• pip3 install -r requirements.txt
• Make a new directory and copy/clone app folder
• In the app directory run flask run (may need to move the wsgi.py file out of app folder)

Or with Docker

• In app folder run
• docker build -t protolith-app .
• docker run -dp 8000:8000 protolith-app
• The app will then be available via web browser at localhost:8000

Attribution and training data

This predictor is a reformulation of the original work published by Hasterok et al 2019. Chemical identification of metamorphic protoliths using machine learning methods. Computers and Geosciences. 132, 56-68

The source data for training can be found at: https://zenodo.org/record/2586461#.XvwL8cgzZPY

Project Organization

├── README.md 
├── notebooks
│   ├── model_assessment.ipynb      
│   └── train_protolith_model.ipynb
└── app               
    ├── requirements.txt   
    ├── templates           
    │   └── index.html
    ├── static           
    ├── model         <- App helper functions and final trained model
    │   │                 
    │   ├── dataHelperFunctions.py 
    |   ├── predict_model.py
    │   └── model
    ├── Dockerfile
    ├── wsgi.py <- Gunicorn wsgi server
    └── app.py  <- Flask protolith predictor app script
       

---