Find Posts By Topic

Facebook hackathon applied Machine Learning to our Open Data platform

In October, Facebook held a civic hackathon, inviting in-house engineers as well as representatives from other tech companies to apply machine learning to our open data platform ( to solve problems.

(If you don’t know what machine learning is, here is a great primer: In machine learning, computers apply statistical learning techniques to automatically identify patterns in data. These techniques can be used to make highly accurate predictions, and these are the techniques that the hackathon participants applied to our open data platform).

Two projects were deemed to be the winning efforts, one focused on issues related to parking and another focused on helping the public get contractor estimates when they want to undertake a major construction or remodeling project.

On Thursday October 26th, the two winning teams were invited to speak at the monthly “Breakfast of Champions” meeting, and share their project ideas and prototypes with over 40 City staff who act as Open Data Champions across all City departments. The aim was to spur new ideas for future uses of machine learning within the City and for the relevant departments to connect directly with the winning teams.

Find n' Park team

Find n’ Park team

The first project, “Find ‘n Park”, tackled the problem of Seattleites spending a lot of time searching for open parking spaces. Seattle is ranked as one of the hardest places to find a parking spot amongst large US cities; and currently there is no way to get good data on the availability of parking in lots or for on-street parking. Find ‘n Park used deep learning vision models to determine how many cars are currently parked in a particular lot, to give real-time availability of parking. You can find more details on this project here:

The second project, “Contractor 5”, tackled the problem of looking for a building contractor and quickly getting a realistic cost estimate for new construction or remodels of existing properties. The Contractor 5 tool models the estimated price to complete a project by to within $5,000 by leveraging City permitting (open) data and using natural language processing to compare your project description with similar projects. It greatly simplifies getting an estimate and increases market transparency. You can find more details on this project here:

Contractor 5 team

Contractor 5 team

Both presentations evoked many questions and follow up interactions between City staff and the winning teams, and gave our Open Data Champs a great insight into the power of ML and how powerful applications could be quickly developed using the open data they are providing. These presentations also provided us with a lot of food for thought re: our 2018 Open Data Plan, especially when it comes to potential investments in using ML to power applications and services, or powering AI scenarios.

Special thanks to the Facebook staff who helped organize this event – Aria Haghighi, Lindsay Amos and SarahBeth Donaghy. We hope this is the first of many partnership interactions!