Boardgame Geek Rating predictor

Introduction

This is a Django REST server built to serve a Naive Bayes model trained to predict the potential rating for a given review about a board game.

The dataset used to train the model is obtained from Kaggle thanks to @jvanelteren

How it's implenented

At it's core, this REST API when invoked does the following.

1. Sanitize the input text from request
2. Apply Stemming and Lemmatizationto get rid of some unwanted information from the text
3. Serves a model that is pre-trained and pickled in the model directory. scikit-learn is used to train, tune, evaluate the model and predict rating.

Instructions

Running a local server

$ git clone https://github.com/iharshadev/bgreviews
$ cd bgreviews
$ pip install -r requirements.txt
$ python manage.py runserver

Deploy to Heroku

Links

1. Detailed explanation on how the model is trained can be found here
2. Live Demo
3. Video Demo