HackMIT 2019
MPGreen
During HackMIT 2019, my team created a data visualization to help make changes to carbon emissions on an individual and government level. We processed and displayed information about fuel efficiency in different cities and under different driving behaviors. This project won Kensho's Data Usability Challenge.
Github RepositoryMotivation
Many people are motivated to help prevent pollution and reduce carbon emissions, but it's not always clear what the best way to do this is. Navigation apps provide options for the fastest or cheapest routes, but not the most eco-friendly routes. Our goal was to give both average citizens and the government the information to make eco-conscious decisions.
An example of what a finished product would look like that suggests eco-friendly routes (route above not necessarily more eco-friendly)
Accomplishments Summary
We processed and created a visualization for a dataset provided by OpenXC to help determine the important factors that contribute to fuel efficiency. We measured and graphed the speed and miles per gallon for different driving behaviors, and for locations in different cities. This would allow an individual make decisions about their driving, and cities to make decisions about traffic patterns and layouts to help reduce carbon emissions.
Technical Challenges
Our data was first processed in python, then visualized using javascript. The raw data came from car sensors that were sampled at different times, so comparing two different sensor values (like odomoter and fuel, to get instantaneous miles per gallon) required interpolation. When a car filled its gas tank, the fuel consumed measured by the sensors would drop, and we had to ignore these drops when calculating miles per gallon. Between the different data sets, the names of sensor values varied ('odometer' vs 'fine_odometer_since_restart', for example). There were also general problems with noisy data that had to be fixed.
My Responsibilities
Creating graphing utility functions for the frontend
Setting up pages and routing for city and behavior data on the frontend
Pull data from OpenXC files and calculate average MPG