It’s [one of] the most wonderful times of year!
I do truly love this time of year. What time? Iron Viz feeder season. A series of three ‘feeder’ contests to find this year’s contestants at the annual Iron Viz competition in Las Vegas during Tableau’s annual conference.
So Tableau 10.2 came out a couple weeks ago with great anticipation. One of the many new features added to this release is the addition of the new spatial file connector. Read about it and more here.
It should be noted now that prior to this viz, I’ve had no experience with spatial files.
So the hunt was on for a great spatial file dataset for me to play with. I found a couple recommendations of great resources to check out from a Tableau Public blog post which mentioned that the city of Philadelphia is doing some amazing things with spatial files through their open data initiative. They weren’t lying. I quickly found a spatial file of 10 years of crime data. I was hooked, it was settled. This is my dataset.
So I downloaded the .geojson file and fired up Tableau. Connecting and getting started with an out-of-the-box spatial file is pretty simple in Tableau. Just as simple as connecting any other file type.
After playing around with it for a while, it became apparent that I was going to need to join some other file(s).
You see, the file I downloaded was each crime of the past 10 years at a lat/lon point. But I noticed that one of the accompanying fields was what they called “district” which is the equivalent to precinct. So I went back to Philly’s Open Data portal and sure enough, they had a spatial file to for police districts. Now all I had to do was join that file on district from the master file in Tableau. I feel like I lucked out a bit after reading some of the twitter conversations and other contestant’s blog entry here and here…and here too. It seemed there were some who needed to use GIS software to merge several different spatial files together.
So back in Tableau, I’ve got crime data rolled up to a district level and down to a street location area. The master file also had what type of crime was committed. The story was starting to come together.I wanted to focus on crime in the different precincts to see how the parts contribute to the whole.
So I barreled through making several different views. I made a trellis chart of % change by precinct…I made a bump chart of total crimes by precinct…I made a bar chart of crime types…I made a bar chart of the most dangerous street corners. Then of course I had the district map and a street corner location map.
More than anything, this project solidified my desire for two future Tableau enhancements:
- Drill down Maps
- Similar to how hierarchies work currently (Year->quarter->month, etc) but with maps and all in the same view. I’ve heard its coming
- Transparent sheets
- We all know how much we all desire this one.
So I had all these elements and I had this idea for a stylish infographic style with the district map in the middle and the others surrounding it. I even managed to “hack” together what appeared to be a zoom-in box that appeared to be zooming in from district to street corner. This is what it looked like.
Looks pretty stylish right? RIGHT?! Yeah that zoom box? That’s the power of annotating points with a single edge text box and floating the other box inside the borders.
I thought it looked really cool. But then…the more I looked at it, the more I wasn’t crazy about it, but couldn’t pinpoint what. So I put it down and walked away from it for a day or so. When I came back to it, I still wasn’t crazy about it. So I floated this image to a few colleagues at work who liked it but all mentioned that it was too busy.
I also sent this image to Curtis Harris, last year’s IronViz winner and all-around great guy. He said, “It looks cool and I like it but…where’s the story?”
YES! This was it, I didn’t have a story with this one in this current layout but we both agreed that I had all the elements TO MAKE a good story, I just needed to refine it. He gave me some suggestions and pointers and I ended up with something that I think both looks appealing but also tells a great story!
One of Curtis’ points was about the title “Crime is on the Decline” He said when he read it, the first thing he thought was the crimes themselves. So I put that first with a classic slope graph. This allows the user to see which crimes have decreased the most then move on to precinct.
I’ve recently learned the key a good data visualization in this particular arena and really a good rule of thumb overall is, “It should look as good as a still image and also an interactive dashboard.”
I’m excited about this entry and to see all the other awesome entries to come from everyone else. Tableau’s user community truly is one of a kind.
Until next time!