Big Data Goes Beyond Business: Town hall meets technology
12 November 2012

by tmartin on November 13, 2012

RIGHT AFTER AN ELECTION CYCLE AND all its attendant cynicism and mudslinging, I find myself needing something to counter all the muck. Nothing like a nice techy conference to clear the air, right?

Last week’s Massachusetts Digital Summit provided blissful geek-speak – and reminded me how much good work actually happens in town hall and beyond.

The annual event, sponsored by the Center for Digital Government ( and a fleet of vendor-exhibitors, brings together the state’s CIO and CTO, agency technology leaders, municipal CIOs, and others working in deploying technology within the state, cities, and towns.

Spend a few minutes listening and you suddenly realize that the commonalities across city and town and state – regardless of size or geography — loom larger than all of the collective perceived “uniqueness.”

A couple seconds later it really hits you: this group not only embraces strategic use of technology, but they also use it every day and you’ve probably benefited from their work.

The range of activity spans from the pragmatic to the cutting edge. Operational functions — like tracking vehicle fleets and putting up web portals – share conversation space with more exotic applications.  You know, stuff like predictive analytics and mobile apps for citizen engagement.

Data – the visualization of data, the moving of data, the use of data, asking questions of data – reemerged over and over in different sessions, mirroring private sector trend lines.

Lots of what we’ve all read about this sizzlin’ hot topic takes the form of Target selling more diapers by clever data mining and analysis to the nth degree.

Or Facebook delivering disturbingly targeted product messages when a user logs on.

Or even political statistical soothsayer Nate Silver ( predicting outcome by running algorithms on the election.

Technically fascinating, and great potential for marketers and business bottom lines and pundits. But sometimes, don’t you wonder … is that all there is?

Good thing I attended  Curt Savoie’s workshop, because now I can clearly say: yes, there is more, a lot more. Savoie, principal data scientist with the City of Boston, started with an overview of trends. Hot topics?

  • NoSQL servers, like Mongo, which works with unstructured data, providing a new lens for telling the story.
  • The job title “data scientist,” which describes someone working at the intersection of math, policy, data, and technology.
  • Descriptive analytics, which uses data analysis to help us understand where we are now.
  • Predictive analytics, which uses data analysis to help us understand where we might be next.
  • Knowledge discovery systems, which let mere mortals ask questions of data in real time and receive meaningful presentation of the results

Most of us lack a Target-sized corporate budget, but it turns out you don’t actually need one to take on projects that yield real results. Shareware tools fill Savoie’s kit:

For statistical and analytical tools:

For data mining:

For data acquisition:

For storage and processing:

Translation: it doesn’t take zillions of dollars to get results, especially if you start small and focused, and walk in baby steps to basic results.

For example, New York City tackled the safety issue of finding illegal overcrowding and apartments within apartments by creative thinking about data combined with a ginormous Excel spreadsheet. The result? Understaffed inspection agencies drew on the results to quickly and efficiently find the most dangerous buildings.

Louisville partnered with United Health to put data sensors on inhalers and has been mapping asthma puffs with time and location data to find (and fix) public health patterns.

Boston took on 911 calls and anticipatory staffing, creating predictive models based on multiple data layers. By taking a data approach, and presenting it in visual and understandable ways, the city began to better understand the ebb and flow of 911 calls and to manage resources more efficiently.

Savoie says he uses five guidelines to put data to work – guidelines that apply in any kind of enterprise.

First, make data easy to understand. Often this requires turning numbers into visuals that tell the story. With a little visualization help, a budget spreadsheet becomes a graphical grid, snapping highlights into key focus.

Second, remember that ancillary data can help tell the story. The usual suspects from the usual sources might be just one part of what you need to know. When he worked on the 911 calls project, Savoie mapped the Boston University school calendar over the Allston-Brighton 911 volume data and suddenly a clear pattern arose.

Third, know that volume doesn’t tell the story. Counting “how many” isn’t enough – you’ve got to put those number in context with other data.

Next, look for patterns that lie within the network. The way that data connects within itself often reveals how things really work.

Finally, never forget that correlation is not causation. All too often people mis-read data and connect unrelated factors. Just because X and Y happen at the same time, doe not mean that one must cause the other.

(If that were true we’d all be solving global warming in our pirate gear, like the gone-viral satire says!

The results so far have literally changed lives – and we’ve only begun to overcome the challenges of bringing together data from silos, building a common language of data, and learning how to ask questions of data. It smells like we’re on the cusp of something potentially transforming.

Hmm, I don’t know about you, but somehow the excitement of selling more baby wipes or getting someone to download the latest Angry Birds just shrivels in comparison to reducing crime, keeping streets cleaned, improving public health, and making our communities run a little better for us all.

If that’s not an antidote to mudslinging political ads, I don’t know what is.

Leave a Comment

{ 1 trackback }

Previous post:

Next post: