Editor’s Note: We welcome a new regular feature and contributor here today. The MediaShift Innovation Spotlight will look in-depth at one great mash-up, database, mapping project or multimedia story that combines technology and journalism in useful ways. These projects can be at major newspaper or broadcast sites, or independent news sites or blogs. Web journalist extraordinaire Megan Taylor will be your guide to these regular bi-weekly spotlights.
What It Is
St. Petersburg Times’ Neighborhood Watch is a database application that tracks weekly house sales in Pinellas and Pasco counties, Florida. Readers can search for home sales by county, ZIP code or neighborhood. Median price and sale count trends are tracked and graphed at one year, six month, three month, and one month intervals. On a neighborhood level, the site plots geographical data on Google Maps and suggests listings to prospective buyers by ZIP code. The application also generates neighborhood-by-neighborhood trend stories by querying the database. The Times plans to expand Neighborhood Watch to cover more counties in the future.
Why It’s Innovative
Every paper has to do these kinds of real estate stories once or twice a year: The housing market has gone up, it’s gone down, it stayed the same, etc. It’s not big journalism, but it is important to the community and it takes a lot of time and resources to put together individual stories for different neighborhoods.
Neighborhood Watch not only provides weekly data on the housing market, but includes an instant story for each neighborhood — a computer program analyzes sale trends to generate a short synopsis of a neighborhood’s market.
This frees up real estate reporters to focus on bigger stories with context and depth. Given the current state of the market, freeing up a reporter’s time to work on big stories is becoming more and more important.
The data is even appearing in the print St. Petersburg Times neighborhood sections; the paper has begun to reverse publish information that originally appeared on the web.
Who’s Behind It
Matt Waite, the St. Petersburg Times News Technologist, is the brains behind Neighborhood Watch.
A Django evangelist and data hound, Waite worked as a reporter for the St. Petersburg Times for almost eight years. He is also the mastermind behind PolitiFact, a popular site where statements from U.S. presidential candidates were fact-checked and rated (including the “Pants on Fire” logo for worst offenders).
In 2004, Waite created maps to compare prices while he was house-hunting; those maps eventually became the seed for Neighborhood Watch.
Waite explained that even though newspapers only have the time and resources to cover broad, flashy stories, it was really the small, local details that interested readers. The same is true in regards to real estate stories:
I give this speech at various journalism conferences about crime. There are two crimes I care about: There’s the crazy dude with the machete who hacks his ex-girlfriend’s new boyfriend’s head off and mounts it to his car and waits for the police to show up; and my neighbor’s lawn mower getting stolen out of his garage.
One of those you’ll find in the pages of the newspaper, guaranteed, the other is the opposite: you’ll never, ever, in a million years read about my neighbor’s lawn mower getting stolen out of his garage in the pages of the St. Petersburg Times. But I and my other neighbors were very interested when that happened. So the trick is to find a way to deliver that kind of information to people in a compelling fashion that doesn’t involve having to pay a massive army of reporters to cover every single thing that moves.
And the beauty of apps like this is that you might not care, but the guy in the apartment next to yours may REALLY care. But it didn’t cost anything to provide that information to whoever might want it, at whatever scale you want it at.
Listen to Waite talk about the origins of Neighborhood Watch:
Neighborhood Watch is built on GeoDjango, a former branch of Django that has been merged with the framework in the 1.0 release.
The site uses a spatially enabled database back end called PostgreSQL, with an added SQL library called PostGIS.
For the most part, the city’s neighborhood associated boundary files provided good neighborhood maps. But in the suburbs, there are developer boundaries, or common knowledge boundaries. In some places, boundaries and names were invented by the St. Petersburg Times.
The data comes from an FTP fetch through the property appraisers.
Waite started building Neighborhood Watch using the same technology he used for PolitiFact — Django .96 and MySQL. But there was a problem: MySQL didn’t support the kind of location relationships he needed.
“I had to establish a relationship between a neighborhood and a parcel,” he said, “There’s no real good way to do that in MySQL, so I resorted to how I did it originally: in a desktop GIS program.”
This had its own problems: Updating.
“Every time I wanted to update the parcel layer, I’d have to recreate that relationship between parcel and neighborhood — it’s called a spatial join,” he said. “If you’re going to be updating something every week, that’s bad.”
So Waite started playing with GeoDjango, “which brings the functionality of a world class GIS system into the Django web framework. With GeoDjango, data models can contain geographic data, i.e. data becomes aware of space.”
Have you received any complaints about the way you drew up neighborhoods?
Matt Waite: Funny thing, no, we haven’t in the areas where we made them up. I’ve received two complaints about the maps. One was that we left out a neighborhood in an area — and they were right, I’m adding it in — and one was that the city neighborhood association map was just one parcel too far over the line in one small neighborhood. So it wasn’t even my doing.
What was your favorite part of Neighborhood Watch to build?
Waite: The geographic components, without a doubt. Location and geography — where you are, where you live — is such a powerful prism for people to look at something through. It makes the data you have so much more visceral to someone when you can show them a map of their neighborhood or their house and how it relates to what you’re showing them. Being able to do that efficiently and powerfully now on the server side is something very cool that I haven’t even gotten my head all the way around yet.
So what has the response been like from people in the newsroom and people in the community?
Waite: Very positive on both accounts. I’ve mostly paid attention to the public’s reaction to it and it’s been very interesting. The bulk of the email I get about it is, ‘We want more, more stats, more measures, more this, more that.’ I’ve gotten a handful of emails from realtors who were mad that I included this sale and not that one, or one of their competitor’s sales but not theirs. Every time, it’s been because I screen out all but the most basic single-family house sales and toss all the rest. I learned with PolitiFact that the best marketing you can do is to answer your email, good or bad, and I’ve tried to do that.
Far and away, the most common comment I get is ‘Where are the condos?’ And the answer is ‘We’re working on it.’ This version only deals with single family houses. The condo market and the house market are two very different things, and we’re going to treat them that way.
Call for More Spotlights
The MediaShift Innovation Spotlight will run every other week. Please let me know of any innovative projects you are working on or have seen lately. It doesn’t have to be from a major newspaper, it just has to be an innovative blend of journalism and technology. Please e-mail me at mtaylor[at]megantaylor[dot]org to submit a Spotlight recommendation.
Megan Taylor is a web journalist whose work focuses on combining traditional and computer-assisted information-gathering with multimedia production to create news packages online. Megan tells stories in English, HTML/CSS/, ActionScript, PHP, photos, video and audio, and blogs at her personal site.
MediaShift Spotlight Innovation illustration by Omar Lee for MediaShift.