The main purpose of this project is to build a web app for a machine learning model and deploy it into a cloud service so that we can build an API to get retrieve the predictions. The ML problem we consider in this test is the sentiment analysis for movie reviews (binary classification: positive or negative).
The dataset is movie reivews extracted from Imdb and can be found at here. In this dataset, we have 1000 positive reviews and 1000 negative reviews
The classifier is based on Support Vector Machine and features are extracted by Tf-Idf. The modeling process can be found in the file ./modeling/ml_model.py. The optimal classifier is saved to the binary file model.file.
The API is defined in the file ./modeling/views.py and the url is defined in ./modeling/urls.py
The optimal model is saved to binary file before being deployed. Firstly, we use Django to deploy it to a local machine to test the API. For example, we can call the API for a review as follows:
- Lauch the server on localhost
python manage.py runserver - Test the API http://localhost:8000/model/api/predict/?review=This%is%a%great%movie.
Heroku is platform that allows us to deploy our web application quickly. After creating a Heroku account and installing Heroku CLI, open terminal to set up the login credentials and create a remote.
- Create a remote for heroku git
heroku create - We can check the new remote by
git remote -v - Push the web app to Heroku service
git push heroku master