This is a rambling account of how I experienced the Tableau Conference 2014 (TCC14) held in Seattle. I will highlight the sessions as well as the city attractions attended adding color where necessary.
Up at 4am, is there another 4? Why wasn’t I informed of this?! It started good, if you are leaving from anywhere, Bloomington airport is one of the best to go from, from parking to the terminal takes less than 30 minutes with security being a breeze. The only place better is from Charlottesville, VA. What makes that better is various historical artifacts at the terminal to peruse, otherwise they are similar.
One thing I’d change about the travel arrangements was the car. I rented a Kia Sorrento. A decent upgrade to my ride, an a upgrade to my Hyundai Elantra Touring, but was unnecessary. The public transportation in the city is truly excellent. I suspect the train from SeaTac to Seattle proper would only take 20 minutes and cost a few dollars.
Once checked in the Motif hotel, the car stayed parked as it was too expensive and a pain to drive around. The hotel itself was good, living up to its name with weird colorful rooms and an open lobby with furniture chosen more for its modernist style than comfort.
After dropping my luggage off at the hotel, some of my collogues and I went down to the waterfront and at ate at The Crab Pot, a local tourist trap. While the salmon wrap was the worst food I ate during the stay, it still was tasty enough, I’d not recommend it for the price and so-so atmosphere.
Then to check-in at the conference.
Unfortunately, this initial start of the conference sucked.
Going up the escalator, the group I was with was treated to dub-step played by a DJ via his MacBook and Tableau employees milling around giving away stickers and other flair to adorn your clothing. The check-in process was simple overall just walk up the some Apple tablets, enter your info and the cleric finished the process, 5 minutes tops. The best part was the free self-styled Tableau backpack, which came in handy as when traveling the conference center and city itself.
The main problem during this process was receiving the lanyard, which as the size of a 7 inch tablet, which made me feed dorky when wearing it outside the conference. When you met someone for the first time, the lanyard caused them to stare at your navel. My eye are up here, HERE! Because they had RFID chips, vendors and those letting you into session poked your belly with wands. I made me feel like to Poppin-in-Fresh Doughboy – he-he.
Session #1: First Timers; Field Guide to the Tableau Conference
Uh, painful and by far the worst session I attended in the four days. Matt Francis and Emily Kund did their best to pack in 10 minutes of content in an hour of presentation. They tried and failed to explain what we were about to embark on in hipster-isque fashion. Upon refection, their presentation style made sense, but just was poorly executed.
Takeaway: If you have been to any conference in your life, skip it if your going to TCc15 in Las Vegas.
I does a choice of a network event with free beer and Seattle. Of course I chose Seattle since I am uncomfortable in large unstructured gatherings. So off to Pike Place Market, which I found that many of the shops were closed since it was after 5. So a group of colleagues and II walked to Etta’s, a Tom Douglas dinning establishment. The overall décor make me feel out of place. I’m much more confortable in jean wearing family establishments. Despite this I had an excellent salmon dish decked out in local farm grown vegetables and their specialty rub. Included was some of the best calamari I have ever ate. I wouldn’t eat there again, but it was a good experience overall.
The first day we were treated to an excellent buffet-style breakfast instead of the usual carb filled donuts and coffee style spread, which I see the norm. For those that didn’t like the fare of egg, meat, potato, and bagel combo they could go to the juice bar or refreshment stations place strategically around the center.
Source: Tabeauza via Twitter
Session #2: Keynote: Christian Chabot (CEO) & Chris Stolte (CDO) – The Art of Analytics
Tableau really came out the of the gate with a wow factor similar to an Apple product launch. It was a high-energy affair that seemed to enrapture the audience. It is not often that you see a CEO of a billion+ dollar company start off with a marijuana joke to get the crowd into what he’s saying.
While topics at the sessions start were interesting, it was a typical affair about how visual analytics was going to change the world and how Tableaus power users will make it happen, so I won’t going in details. You can go to the conference site and watch. The more interesting things were about upcoming features.
Terminology explanation is in order before going into the new features:
Pills – These are the measures and dimensions when used in a visualization
Shelf – Anything that holds pills such as Pages, Filters, Marks, Columns and Rows.
- Enter measures on shelves. This bypasses the Create Calculation popup. You can save them as measures too by dragging them into the Measures area.
- Auto complete on the shelves. This is similar to what Excel 2010+ does when you enter a formula into a cell on the taskbar. Should be a great feature that will make simple calculations easier to enter.
- Analytic drag and drop. Users will be able to drag, say Median pill onto a measure on a shelf and change the calculation on the fly.
Takeaway: Expert analysts will enjoy them and the changes will make it easier for newer users to build out visualizations.
- Search – Users will be able to type in say, Colorado, and have the map zoom into the correct location. It will work with countries, states, counties and cities (and maybe more)
- Circular and Lasso selection – Users will be able to select areas in anyway they choose. Right now you can only select in squares.
Takeaway: Great updates for power users, but casual users will not notice these feature. They are not prominent on the UI.
- Caching will make maps update quicker
- Maps will pre-render in the background so there is no lag when moving around.
- Calculation will now use more than one core. Most systems have at least 2 to 4 so expect much faster results.
- Calculations will persist once calculated. This will help when opening a book to continue to work or switching between dashboards.
Takeaway: Minor but useful improvements.
- Auto clean up Excel imports. Can’t wait for this feature. I will no longer have to format Business Objects reports before entering them into Tableau.
- Split on qualifiers. If a column has something like this, Illinois-Bloomington, you will be able to separate it into two columns on import.
- Web data connector. This allow users to pull into content from web APIs. This might be an awesome bridge to import data from Adobe SiteCatalyst or other could data API sources.
Takeaway: Not flashy, but these are the most important improvements. Data cleaning takes up large portions of most analysts time.
- Can control colors on the screen now.
Takeaway: There were other features talked about bit I didn’t write them down. Storytelling is still worthless in an enterprise environment.
- Caching calculations. Same as above
- Infinitely scrollable views. Finally, I never liked that it only displayed 10 workbooks at a time.
- Cleaner interface with more meta data displayed on screen.
- Easier to adjust security features on a per user/group basis.
Takeaway: Features will make finding work easier. The server features should help infrequently updated workbooks run faster.
- “Cloud 2 Cloud” login.
- Push data from one cloud based software (such as Sales Force) to Tableau Cloud.
Takeaway: Only good for those using Tableau’s cloud instead of Tableau Server.
- Caching calculations again.
- Ability to favorite workbooks and to highlight
- Web and mobile app will have more ways to edit. Eventually Tableau wants parity with desktop and mobile operations.
- Snapshot – Ability to save data to use when offline
- Mobile workbook creation
- Ability to pull in Excel files and create dashboards
- Mobile app can zoom down to the record
Takeaway: In 5 years it could be a game changer, but most changes are minor or untested in real world scenarios. I personally like this feature because it means my tablet will have more powerful editing features.
Session takeaway: Tableau says that they will invest more in research in the next 2 years than since the inception of the company. If that is the case, expect great things when these upgrades arrive in the coming months.
Session #3: DATA @ NFLX: Building a culture of Analytics Everywhere at Netflix
Of all sessions on the first day, this was the one I wanted to see and it didn’t disappoint. Blake Irvine and Albert Wong went into detail about the analytic strategy of Netflix. It turns out that they are going down the same path as Country Financial, just on a larger scale.
Each division has its own analytic group. The benefits are that the system engineers can become system experts rather than generalists. This improves analytics expert response time to meet user needs. Each area operates independently and therefore Netflix must own multiple reporting data stores. To minimize duplicate reporting/metrics, analysis teams have frequent knowledge sharing meetings.
My take: This is a good short run strategy as a company grows, but becomes burdensome as it ages. All these process need experts to keep them running. Knowledge is often lost when people move jobs so it becomes harder to maintain and upgrade. Even with open-source technologies, which Netflix often uses, eventually the company will have a difficult time with maintenance and an nearly impossible time switching technologies when the need arises.
Other things of note:
- The company does a lot of A/B testing to improve user experience. Tableau is used in some of these workflows.
- Mostly reporting systems use open source technologies to gather and store data, though they still use Tableau for the front end and occasionally Teradata in the back.
- Nearly all of their data is in the cloud using Hadoop and Amazon S3.
- Have an OE department that uses Tableau through Hadoop and Teradata.
- They are open to developer changes on the fly (empowerment) to reporting systems. Many have access to the underlying database and can change code to meet their users insights need.
- The IT is reluctant to remove reporting systems once built. This is one of the reasons why I think their strategy will become unsupportable in a decade or so as people who build the systems leave.
- There is little to no data governance.
Takeaway: Data, analytic, and insight silos are common no matter if the company in the new or old economy.
Session #4: Visualizing Unstructured Data with Tableau, Featuring Bill Inmon
Honestly, I didn’t get much of out this presentation. Bill was too detailed in his explanations of unstructured data using taxonomy created by Wand. When pivoting to his Verizon and 1st American Tableau reports, I was unimpressed. Having used unstructured data from call center and surveys before, this added nothing someone couldn’t do with R or Rapid Miner’s text analyst add on.
The main problem was there was no sentiment or context analysis, so you have to use qualitative or other data to indicate whether or not groups of calls were positive or negative.
Takeaway: I still think that Tableau stinks for mining/displaying unstructured data. It is better to use a specialized tool and the display the insight in Tableau.
Food provided continues to impress. The conference had a main place to eat near the vendors booths, but also stocked at least a dozen fridges with wraps, drinks, and salads so you never had to go far for vittles. I am sure a few people left for lunch, I did for a bit to purchase some bottled tea, but if you weren’t picky as me, you could eat hardily, even if vegan.
The main problem with lunch was the wait. With 5,500 and at least 1,000 Tableau employees going to each at once (all of the sessions ended at the same time) people could wait a half and hour to get to the buffet.
Session #5: Zen Master Tips & Tricks Panel
In this session, the there were three presenters, Kelly Martin, Craig Bloodworth, and Mark Jackson each with a different style they used to create visualizations. Overall, the session was average, the best part was looking at the examples rather than the tips they offered. This and the following session really helped stoke the design side of my brain.
- There is no need to type if or case statements. Tableau will handle the branching for you. If used, I think this will make statements less clear to others.
- There is an ability to do top ten (or any number) on by a category basis [link.]
- There is an ability to do context filters [link.]
- There are two API for Tableau Server. Version 1 is undocumented and version 2, new to Tableau 8, is documented. To understand version 1, you can use Fiddler to discover it features by spying on the traffic between the computer and server. It also helps to code by using Alteryx to code the API and pull XML data.
- Use blank text boxes over data to prevent people from clicking on it.
- The [+] in front of a dimension in a hierarchy is called the power key.
Takeaway: Tableau is flexible. Most of the tips offered have multiple ways to accomplish the same thing.
During this session, some of the examples helped me figure out ways to solve some of my problem dashboards. During the session, I drew an updated design for the Agency Dashboard. Hopefully, I will have time to integrate it in a few weeks. Note: I know my drawing sucks, bare with me here.
Session #6: 10 Tips to Becoming a Tableau Jedi
I was impressed with Marc Rueter’s presentation. He was knowledgeable and went just at the right speed to explain his tips without having to hurrying like many presenters did with the limited time they had.
- You can drop dimensions and measures on top of existing pills to replace them. The new pills will inherent the previous pills existing formats and display. I use this trick when building dashboards. Frist, I duplicate the worksheet and then make the change on the copy.
- Create a Summary. on the visualization, right-click > Summary
- Create a source average or median. Duplicate data source w/o connecting it to its parent and place the measure the chart. For best results use dual axis. I am not a fan because you have to use data blending and a duplicate source.
- You can create trellises
- You can create drill downs on dashboard using actions (Country > State > County)
The rest were too complex to explain or I already knew they existing so didn’t write them down.
Not knowing the extent of food that would be on hand at the Data Nite Out, I went to eat at Elliott’s Oyster House out on Pier 56. I think it had the best food I ate during my stay. This might be because I didn’t eat much during the conference, but whatever, it was great. Chowing down on their Sockeye salmon ruled and tasty spicy clam chowder while looking out on the harbor was awesome.
Getting back to the hotel, I was soon on a bus to the Seattle Center that Tableau rented for their networking event. This location is familiar to tourists. It contains the Space Needle (which I had already been in on a previous trip) and the starting leg of the monorail built during the 1962 World’s Fair.
I wasn’t prepared for this event’s scope. With 8 to 10 bands and dozens of food and adult beverage locations, it was easily the most impressive part of the entire conference. They did it in style. Each area has a unique theme, food type, and activity. After wandering around sampling every sugar, salt and, grease concoction I could find, Sir-Mix-A-Lot started. I never been a huge fan, but the show was high energy helped in part by the free booze Tableau provided those in attendance.
Ode to Sir-Mix-A-Lot
I like big data and I cannot lie
You other quants can’t deny
A dataset comes in with a little bit of analysis
And a pie-chart in your face
You get insight
Update #1: Spelling and photo sizing changes. Added some graphics based on comments.