ML Tools

ML Tools is a collection of three complementary parts: ML Interface Wizard, ML Measure, and ML Contribute. These tools are designed to simplify the development and deployment of Machine Learning models.

Please note that during the development process, this repository's name will be changed to ML Tools to accurately reflect the scope of the project.

---

ML Interface Wizard

ML Interface Wizard is a full stack application that aims to simplify the creation of front-end interfaces for Machine Learning models. It provides a user-friendly interface to interact with models, check their correctness, and provide additional testing data. The application generates a UI for any Machine Learning model based on model metrics and provided metadata.

Problem Statement

During the development of a data application based on a Machine Learning model, it is often necessary to create a front-end application specifically for that model. However, when the underlying model changes, a new front-end needs to be generated based on updated metrics and metadata. ML Interface Wizard solves this problem by automatically generating a UI for any Machine Learning model.

Tech Stack

Features

• Web-based UI for interacting with Machine Learning models.
• Easy model usage and correctness checking.
• Support for providing additional testing data.
• Automatic UI generation based on model metrics and metadata.
• File and state management for up to 4 possible file uploads.
• Server-side validation for uploaded files and configuration files.
• Context-based state management solution using the React Context API.
• Dynamic generation of UI form elements based on configuration files.
• Support for uploading files in the "pkl" format and configuration files in JSON format.
• RESTful API for communication between the frontend and backend.
• Integration with Uvicorn, Python, FastAPI, and MongoDB on the backend.
• Deployment using Docker, AWS, Linode, and Azure.

Uploading Files to Generate the Form

Getting a Prediction by Filling Out the Form

Deployed Version

You can also access a deployed version at ML Interface Wizard.

Installation and Setup

1. Using docker compose for Backend

docker compose up -d

This will setup the backend application with all necessary dependencies.

Alternatively, look into step 2.

2. Clone the backend repository and run it by following guide for the Docker version:

git clone https://github.com/nikolaDrljaca/interface-wizard-backend.git

# 1. Create virtual environment
python3 -m venv env

# 2. Activate created env
source env/bin/activate # Unix
.\env\Scripts\activate # Windows

# 3. Install packages
pip install -r requirements.txt

# 5. IMPORTANT - Create `.env` file in the same directory as `docker-compose.yml`
This will be used to setup the MongoDB instance. The following keys need to be present.
Define values as desired or leave listed defaults.
DB_USER=test
DB_PASS=test1234
DB_HOST=localhost
DB_PORT=27017

# 4. Setup MongoDB instance, make sure Docker is installed
docker compose up -d

# Controll the docker container
docker compose stop -> stop or shutdown the container, but DON'T tear it down, the data will not stay
docker compose start -> If the container is stopped, start it this way

# 5. Run the app
uvicorn app.main:app --reload

If you are having problems with how to run backend, please check the following repository:

ML Interface Backend

If You Want to Run the Frontend Locally:

1. Clone the frontend repository:

git clone https://github.com/davutkulaksiz/ml-interface-wizard.git

2. Install the required dependencies:

npm install

3. Start the frontend development server:

npm start

4. Access the application by visiting http://localhost:3000 in your web browser.

Contact

For any questions or inquiries, please contact the project maintainers:

• Nikola Drljaca - GitHub
• Davut Kulaksız - GitHub

Thank you for using ML Interface Wizard! We hope it simplifies your Machine Learning model development process.

---

ML Measure

Will be updated...

---

ML Contribute

Will be updated...