DublinBus Project

The DublinBus Project is a web application that estimates journey times for trips on the DublinBus network using predictive data analytics. It was created as part of the research practicum for UCD MSc in Computer Science during the summer of 2018 by Chen Yiming, Thomas Anderson, Rosanna Hanly and Sheena Davitt.

What it does

It allows the user to enter details of a DublinBus route (or subsection of route) and a date and time; the web app will then estimate the journey time and display it on the user interface.

It is currently hosted at www.dublinroute.com

Folder structure

├── backend
├── config
├── data_analytics
├── frontend
├── otherScript
├── tests
├── twitter_alert
└── weatherapi
gitignore.git
readme.md

Backend: code that returns journey estimates, interacts with database, creates logins, etc.

Config: Service files for background processes (eg weather api).

Data Analytics: Code and notebooks related to the data cleaning and model creation stages

Frontend: All code used to create user interface.

Tests: Unit tests were performed using the Python UnitTest module. Each test file can be run with the command: python -m unittest tests/test_something.py

Twitter_alert: code to create React Twitter component for UI display

Weatherapi: Scripts to collect current weather and forecast data and write them to database

Deploying the application from Github

Clone the application from Github

First, clone the Dublin_Bus_Project repo from GitHub

Install environment requirements

1. Install anaconda

2. Create your environment with conda. For example conda create --name myenv

3. Activate the newly created environment: source activate myenv

4. Install requirements.txt using pip. For example pip install -r requirements.txt

Setup the database

1. Create an empty relational database called dublinBus
2. Run the SQL commands in the /config/database.sql file to create the tables

Start backend services

Start the following backend services:

Weather forecast - Obtains the 5 day weather forecast and puts it in the database: systemctl forecast.service start

Weather api - Obtains the current weather and puts it in the database: systemctl weatherapi.service start

Current GTFS schedule - Starts a SystemD timer so that a new GTFS schedule will be downloaded and uploaded to the database on a weekly basis:

systemctl current_schedule.timer start

Prediction model

This section descibes how to obtain the prediction model files in .pickle format. Processing of the files can take a significant amount of time.

Prepare the files: systemctl bus_extract.service start

• This SystemD service executes /data_analytics/scripts/job master.sh which executes /data_analytics/scripts/extract_bus_line.sh for each bus line.

Train the neural network prediction model for each bus line python /datanalytics/prediction_model/run.py

Note: Many of the files in this project require passwords in order to successfully run. Enter your login info where needed (eg in the dbconnect.py file in the weatherapi directory)

Starting the web-server

To start serving the application, the Nginx and uWSGI servers needs to be started

1. Enter the command:

   systemctl nginx start

2. Enter the command:

   uwsgi /backend/uwsgi.ini