8/19/2015
Tableau Conflict of TUGs
The outcomes are in from the 2015 Conflict of Tableau Consumer Teams (TUG), here, and the Cincinnati TUG received! If you have not seen the gathering of ultimate visualizations, you may see all of them here.
I assumed it is likely to be attention-grabbing to stroll by means of our course of and among the challenges.
Background
When the Conflict of TUGs was first introduced I offered the thought at certainly one of our Cincinnati Tableau Consumer Group conferences. We had very robust attendance at that exact assembly, with greater than 70 folks in attendance. After I inquired about their curiosity in taking part within the Conflict of TUGs, greater than half of the folks within the room confirmed curiosity in taking part. So Russell Spangler, certainly one of our group leaders, registered the group for the occasion.
The principles
As a bunch, we wanted to get collectively for our July assembly, regardless of the day, and obtain the information and full the viz throughout that point. We made an effort by means of our group web page and through some emails to slender down a date. Sadly, we did not get a big response and the responses that we did get have been conflicting. Being July, in the midst of summer season, folks have been out of city or had different commitments. We settled on one of many dates, however as we bought nearer to the date we misplaced a couple of extra folks.
Go or No-Go
Now got here a alternative, can we transfer ahead as deliberate? or can we merely again out due to the low response and the conflicts of dates? We even provided to maneuver the assembly knowledge, however time was working out and we did not have that many choices. We seen a video that was posted on-line that the Seattle TUG posted and noticed that the information was on Ohio colleges. So we agreed 1.) we already dedicated to doing this and a pair of.) the information is in our yard, so we should always take part.
Day of the competitors
We arrange a shared Dropbox folder and Russell downloaded the information early within the morning and had all of it able to go. Hamed and I deliberate out a method and shortly after midday we have been off to the races. I began working with the information, which was efficiency knowledge for the Ohio colleges. I pointed Hamed to Tableau Mapping BI, here, the place Tableau Zen Grasp Allan Walker had posted a file of the US Faculty Districts. We had college district as a subject within the file, but it surely wanted to be matched as much as join the 2 knowledge sources. Hamed started that processing of matching, which additionally required some knowledge cleansing because of many unhealthy polygon factors. I bought and downloaded some photographs of a chalkboard and all of us mentioned the theme and agreed on the idea. I created some starter photographs for the entrance display and the dashboard pages and located some cool chalk fonts. Russell labored on some design components, making titles photographs from the customized fonts and creating some social media icons, and so forth. In the meantime, I began exploring the information.
Hitting a brick wall
All of us run into challenges with the information. Knowledge isn’t good and in reality, I inform my college students, « By no means belief the information ». (Michael Wu prefers me to say « By no means belief the belief of the information », which can be extra on level.) About half-hour into some knowledge exploration and making a bunch of charts to make use of as the premise for the visualization, I hit a brick wall. After I began mapping the faculties, I found that the information was not appropriate. The college addresses, metropolis, state and latitude/longitude weren’t accurately lined up with the college info. Every college had quite a few columns of efficiency knowledge, however nothing else was lining up. Colleges exhibiting an tackle in Cincinnati have been exhibiting up in a faculty district in Cleveland. I used to be in a position to find an information supply from the State of Ohio, so I used to be in a position confirm that the efficiency knowledge was appropriate. What appeared to have occurred was unhealthy geocoding. It seemed just like the rows have been displaced from each other within the geocoding course of.
I referred to as Russell and Hamed and we talked about it. Possibility 1.) Bail, and simply notify Shawn Wallwork that the information was unhealthy and we could not proceed. 2.) Work with the unhealthy knowledge as finest we may making visualizations with out college location or simply permitting unhealthy knowledge or 3.) attempt to repair the information. I went again to the file from the State of Ohio and was in a position to append a brand new tackle (i.e. the proper tackle) for every college. Then I started geocoding. I usually use FindLatitudeandLongitude.com, but it surely was having some bother and saved timing out on me, so I used one other device that I actually like, Geocodio to geocode the file. The file correction and geocoding actually set us again, taking time to construct a brand new file and geocode it. We determined we’d get collectively at my home round 6pm and begin over with the brand new knowledge. That gave use 4 hours.
Viz Time
Hamed arrived round 6:30pm and we bought began with the recent, corrected file. He started matching once more with the Faculty District file. In the meantime, I began placing collectively the visualization with the proper knowledge. We knew we needed to make use of college district and plot colleges, so it actually was vital to have that corrected knowledge.
We labored up till the wire, submitting the visualization to Shawn at 11:57pm. We weren’t accomplished, however we have been out of time, having misplaced a very good little bit of time coping with the information drawback and having to redo some issues. All in all, we have been fairly near the place we needed to be. We did return and add the ending touches to the visualization. We additionally needed to construct some extra search performance for Faculty District and Faculty, perhaps textual content field looking, however we simply did not have the time to get these in.
Classes discovered
By no means belief the information. When taking a look at a brand new knowledge set, it is at all times vital to go searching, verify the information. Take a look at min, max, vary, nulls, look at the size. On the whole, simply poke round. On this explicit case, as quickly as I mapped the faculties I may see there was an issue with the information. When filtering « Cincinnati » college district and there are colleges exhibiting all around the state of Ohio, that was an enormous learn flag. My quick thought was that I did one thing improper, a nasty filter or the improper subject, however after I went again to look at the information I may rapidly see there was a problem with teh underlining knowledge.
Do not quit. We had drawback after drawback. Conflicting schedules, folks dropping out, unhealthy knowledge points that we wanted to resolve, goecoding issues and a ticking time clock. We saved urgent ahead and on the finish of the day we have been actually happy with what we had achieved in such a brief time period. It was an excellent alternative to compete with some nice Tableau Consumer Teams internationally and we had enjoyable doing it.
This is our last viz with a couple of added touches.
As at all times, be happy to electronic mail me at Jeff@DataPlusScience.com
Jeffrey A. Shaffer
Observe on Twitter @HighVizAbility
