paulbenge/VintilaSiFormatia

MPS project

VintilaSiFormatia

Table of Contents

• About the Project
  • Used Technologies
  • Wiki webpage
  • How does it work?
• Getting Started
  • Prerequisites
    • Backend Prerequisites
    • Frontend Prerequisites
  • Installation
  • How to run
    • Training and evaluation
      • Training
      • Evaluation
    • Overall Application
      • Backend
      • Frontend
• Contributing
  • The Team

About The Project

This ML project is about inference for patients that are suspects for COVID-19 desease.

Used Technologies

• Python
• Ludwig
• Flask
• Angular

Wiki webpage

https://github.com/criscojocaru/VintilaSiFormatia/wiki

How does it work?

• Python script will read the training data set and preprocess it
• After preprocessing, we will train the model using Ludwig
• The user accesses the website and loads the dataset
• Call from frontend towards backend is made to evaluate the dataset on the trained model
• Results are returned and showed in an easy to understand way

Getting Started

Prerequisites

Backend Prerequisites

1. Install Python 3

2. Run the following commands in an elevated PowerShell window / CMD window:

python -m pip install --upgrade pip

pip install ludwig

3. (Optional step) If you're using Windows, on build 2004 you have to run:

pip install numpy==1.19.3

4. Also run the following commands:

python -m spacy download en

pip install flask

Frontend Prerequisites

1. Install Node.js, LTS. Make sure to also install npm and select Add to PATH

Installation

1. Navigate to /sources/Angular_UI in a PowerShell elevated prompt

2. Run the following command:

npm ci

How to run

Training and evaluation

Training

• python .\sources\main.py train .\input\training.dataset.csv .\output\preprocessed.dataset.csv .\sources\model_definition_train.yaml

Evaluation

• python .\sources\main.py evaluate .\results\covid_inference_train\model .\input\testing.dataset.csv .\output\preprocessed-exp.dataset.csv .\sources\model_definition_evaluate.yaml

Overall application

Backend

1. Navigate to the project's root in a PowerShell prompt

2. Run the following command:

python sources/server.py

Frontend

1. Navigate to the /sources/Angular_UI folder in a PowerShell prompt

2. Run the following command:

npm start

3. Access the website application at the link: http://localhost:4200/covid-inference

Contributing

The team

1. Bengescu Paul, 341C5, Developer
2. Ciudin Andreea-Alexandra, 341C5, Developer
3. Cojocaru Cristina-Gabriela, 342C3, Team Leader
4. Petrescu Ioana-Raluca, 341C5, Analyst
5. Vintila Marius-Mihai, 341C5, Project Manager