delarroqua/velib_predictions

Availability predictions of Velib stations in Paris

Vélib' Prediction App

Predict availability of bikes stations in Paris. You can find a series of article on these links

• [Part I: Data retrieval and storage in AWS] (https://medium.com/@pierredelarroqua/v%C3%A9lib-in-paris-part-i-discover-aws-and-postgresql-e0cf19d09988#.1fki9slsc)
• Part II: Web App with Flask that use a simple model in Python to predict availability of the stations
• Part III: Improving the model with additional features and a better algorithm

Installation

Dependencies

• Python 3
• PostgreSQL
• Flask (http://flask.pocoo.org/)
• Requests (http://docs.python-requests.org/)
• db.py (https://github.com/yhat/db.py)

Create conda environment

conda create -n velib python=3.4 anaconda

Switch to velib environment

source activate velib

Install python packages

python setup.py install

Train a Model

python train_model.py

Run the application

python run_server.py