This repo contains a dummy project created during Build@Mercari 2021 to learn about the Deployment of ML Models

Live Demo: http://udemy-mercari-api.herokuapp.com/

Steps to run the application locally after cloning the repo:

1. Make sure you are inside the root directory mercari-item-classification
2. Download dataset from https://www.kaggle.com/c/mercari-price-suggestion-challenge
3. Put the train.tsv and test.csv under mercari_model/datasets/
4. Run python -m venv env to create virtual environment, then activate it source env/Scripts/activate
5. Run pip install -r requirements.txt to install required packages
6. Run tox -e train to train and save the pipeline for later inference
7. Run pytest or tox to run pytest scripts make sure all APIs are working
8. Run export FLASK_APP=flask_apis to set which module contains the Flask API
9. Run flask run to serve the api, or run python wsgi.py
10. Test using any frontend app or Postman to try the APIs

Next todo:

• circleCI for CI/CD