cityofaustin / hack-the-traffic

A Transportation-Themed Hackathon hosted by the City of Austin and the UT Center for Transportation Research

ATX Hack the Traffic

ATX Hack the Traffic is a hackathon series that seeks to use data and technology to solve Austin's transportation challenges. The project is a City of Austin initative carried out in parternship with the UT Center for Transportation Research.

Our next event will be at ATX Hack for Change on June 2nd. Visit our Open Source Traffic Sensor repo for details about the Hack for Change project.

Event Schedule

• Hack the Traffic - March 4, 2017
• ATX Hack for Change - June 2 - 4, 2017
• Transportation Data Challenge - Fall 2017

Talk to Us

Catch up with us on Open Austin's Slack channel #HackTheTraffic.

Email us at [email protected].

hack-the-traffic's Issues

`Open source traffic sensor` link cannot be found

Retaining enough of the MAC address to determine device vendor

First, thanks so much for releasing this data! It's an amazing dataset and I anticipate that we only touched the surface of potential solutions it enables at the event yesterday.

One problem I discussed with a few other people at my table was distinguishing cars from bicycles from pedestrians. We could draw thresholds based on the assumption that cars travel faster than bikes, which travel faster than pedestrians, but this is not always true: during rush hour bikes might outstrip cars; in rush hour + accident, pedestrians might even win.

While cars / bikes / peds might all be carrying iPhones, there are likely some devices that are clear indicators of a specific category, like a smart dashboard system in a newer car.

For this purpose, it would be useful to retain the first three octets of the original MAC Addresses, which indicates the vendor of the device, and only anonymize the subsequent octets.

• IEEE provides lists of the top-level vendors at https://regauth.standards.ieee.org/standards-ra-web/pub/view.html#registries (for example, my iPhone's Bluetooth MAC address starts with CC:20:E8, which maps to "Apple, Inc.")

Having even limited vendor cues could be useful for bootstrapping a clustering algorithm to automatically distinguish (to some degree) between cars / bikes / peds.

HackTheTrafficBluetoothRadarServer

Th repo is at https://github.com/shravan2x/HackTheTrafficBluetoothRadarServer.

Eventbrite sold out

Is it actually all full?

data.austintexas.gov (Socrata) vs. data.world

• Which one should be preferred for correctness / timely updates?
• Are the csv files as available through data.world the 'raw' form that is fed into Socrata and thence only available by the usual Socrata API / export features?

And if relevant, can you briefly describe the ETL process and anticipated frequency of updating those data sources with the latest sensor readings?

MAC address anonymization procedure documentation

Related to #9, I have a few questions about the MAC addresses that appear in the dataset. Can you explain / describe:

• How are the anonymized addresses generated, precisely?
• If I see, for example, 8f:bd:9e:e9:18, in one day's data, will that be the same device as every other 8f:bd:9e:e9:18 in the rest of the data?
  • If not, what are the boundaries of identity?
• Why are the (anonymized) MAC addresses five octets instead of the full six?
• How is the matching performed?
  • If the sensors themselves truncate the last octet, is the matching process going to confuse similar devices? (it's probably rare, given the size of the MAC address space, but possible)

wc description says "lines, words, and lines"

It should probably be "lines, words, and bytes."