Thursday, August 22, 2013

Troubling Time with Tableau

Since moving to the public sector, I've used a lot more software packages than I thought. Prior to my current gig [which you can check out on my linkedin page], I worked for a quasi-tech company. Some would say it was a sales company with a tech focus. Either way, coding was much more open source than my current gig (for better or worse). Over the past few months, I've transformed into a data artist, also known as a Data Visualizer. Although what I do for work is not public, I've done some other public work that I can reference to express my views on software. Ok, enough talking already. Below is the visualization, enjoy and read my rant that follows.

My Goal: As the Atlanta Braves continued to pound the Washington Nationals, I wanted to know when that gap really began. So, I started by plotting the game by game wins. You can see that below or for a better view this link http://public.tableausoftware.com/views/NLeast/Dashboard1?:embed=y&:display_count=yes :


My troubles:

1. Messing with the titles, axis and scale proved to be harder than expected. For some reason you cannot move the column name below the numbers.

2. Making the teams’ logos was also complicated. Not impossible, but it required a hack.

3. It is literally impossible to make this a rolling line graph. I have NO idea why. I searched around and found pseudo-hacks, but for a game to game basis this was actually impossible.

4. I wanted to add more teams; but I thought it would get cluttered. Another problem was that I couldn't group things. I wanted to do hierarchical analysis, where you could compare the NL East, the NL, and the NL wild card. However, for reasons not known to me, I couldn't do this. Apparently Tableau has this feature, but it wouldn't work.

5. The play feature doesn't work on the web. There should be a 'play button' that will moves from game 1 until game 126. This was a huge downer for me. Mainly because I didn't realize the 'Public Version' lacked this functionality.

 My view on Tableau : 

As companies continue to think “big picture”, they need a “big picture”. That is where tableau succeeds. Unlike Excel, Tableau is quite good at handling more than a million records. It can do some pseudo-sql and allows the users to drag and drop data. That being said, Tableau is somewhere between Excel and say D3 (which I love). A person not familiar with programming can use it, but it allows people to devote 40 hours a week on making reports that are palatable. Most proprietary software, Tableau included, is slow to adjust to new trends. Design teams function because they design fast. When branding your company, you need to think abouttomorrow not today. Tableau 7 still struggles to use newer types of visualizations like word clouds and bubble charts. Rumour has it Tableau 8 has these features, but that requires (ughh) a new license.

Stackoverflow is the new F1: 


I had a college professor who never required a textbook, which was simultaneously great and horrible. It was great because I didn't have to spend $300+ on a book that I would likely read once, but it sucked because we did not have reference material. He would always say, “F1 (which is the help button) is your textbook.” When working for a company, they rarely tell you to use F1. So, you generally have a few options. You can ask to the company expert in that software, you can call the software vendor for customer support or you can Google it (which ultimately takes you to stackoverflow). The open community seems to be hurting the software companies. Before the internet people had to find that person, now it’s much easier. With a simple web query, you no longer need customer support, which means you may not need the special software.

2 comments:

V for ..... said...

have you heard of the R package 'Shiny'? what are your thoughts on that?

Joshua Patchus said...

I love shiny actually. The only thing that R cant do is host my server for me, i guess thats the nice thing about tableau, they host the image for you.

I know that R-Shiny is pretty unreliable on a system server. It also has serious security flaws, which tableau apparently has cleared. I love using Shiny for internal purposes, but public facing not as good.