As part of our month long look into government data, we have created a viz that examines graffiti data posted by New York City on their Data Mine. The differences between boroughs is striking: Brooklyn has more graffiti incidents than Queens and Manhattan put together.

Looking at the boroughs by precinct across The Big Apple, one can quickly see that there are some differences in how graffiti is handled. For instance, Staten Island has very little graffiti, but the graffiti they do have lingers without cleanup for almost twice the citywide average. On the other side of the spectrum, Manhattan has over 2000 incidents of graffiti, but it is cleaned up in less than 17 days on average.

다른 유익한 자료도 확인해 보세요


This is nice. A great example of public data use. Someday this sort of public data analysis can be prolific like YouTube.

Thanks Julius, that is exactly what we are trying to instigate with Tableau Public. Thanks for your support!

So, I got to thinking, what if one wanted to examine this data at the incident level, to have more insight. Are at the tags on one side of the Precinct or another? How often is the same address tagged? What about recency -- any gaps precinct by precinct. To get a start on this, I geocoded a sample of the records using; the original data has an address field and a Borough field, but I had to add a State field to do the geocoding and hope for the best since there is no ZIP Code field in the data.

Then, I modified Ross's viz to come up with the following draft:

Rooftop Investigation of Graffiti in NYC

(BTW, any way to embed the actual viz in the comments here, instead of a link out to Tableau Public? I tried using the code generated by Tableau Desktop but it did not work.)

Ross's viz had pie charts at the location of each Precinct house, but if you do some digging with the revised viz, you can see that some Precincts are likely long narrow affairs, others are multi-focal geographically, etc. You can filter or highlight on the Resolution group and Status to see if there is geographic bias within a Precinct for cleaning or not cleaning, leaving open or pending, etc. Lots of ways to drill down and gain insight here. Looking for comments and yeah buts, and improvements, better/quicker ways to do the geocoding, etc.

Peace and All Good!
Michael W Cristiani
Market Intelligence Group, LLC

Very cool! The only thing I noticed that was odd is that selecting a mark in "Time to Close by Resolution" results in the three bottom vizes being cleared. As for posting embedded visualizations in the forums or blog comments, that's not something we've enabled for guests yet.

I hope this is the appropriate place to write this comment - if not, apologies! I'm trying to do something similar to this, except with charter schools in NYC. How did you get the data points in your map graph to be divided to show status? If it helps, I'm trying to show number of students that applied to a particular charter and the percentage that were admitted for each school that is mapped on the graph.

Thanks in advance!


Kim, those are "Pie" marks on the map graph. Selecting that mark type will allow you to specify fields for wedge size and starting angle.

I like it, especially Mr. Cristiani's enhancements. The mapping engine needs some tweaks (zoom out function, satellite image view, etc) to get up to speed with the kind of geographic freedom Google has taught us to expect.

블로그 구독