One of the foundations of the Data School curriculum is the client projects. Essentially, we identify clients that have problems they need solved, dashboards they need built, etc. and the Data Schoolers work on these for a week each. But not a week really. Only about half of their time during client project weeks is spent on the project itself; the other half is training. A few key skills the team develops during these projects are:

• Communication – They have to do all of the communication with the client. From the initial briefing to set up the scope of the project through the handover at the end, they manage it all.
• Project Management – Each DSer gets the chance to be a project lead for a week. This helps them practice leadership skills, coordinating a team, etc.
• Business Analysis – Clients give a brief on the project to the whole team on Monday morning. With the exception of the project lead and me, no one else on the team knows anything about the scope of the project. This means they have to get really good at asking questions, listening, and scoping the work.
• Presenting – Every Friday at 3pm the team presents their work for the week. On client project weeks they present directly to the client, who often has lots of questions along the way. This gets them comfortable with presenting to an unfamiliar audience and being able to answer questions on the fly. This is easily my favorite part of each week.

Recently we had the pleasure of working with Tableau Zen Master Jonathan Drummey on a project for PATH. I asked Jonathan if he’d be willing to share his experience working with DS7, so here’s my interview with him. Thank you Jonathan!!

Andy Kriebel: You recently ran a project with The Data School. What was the project about? How did you come about getting a project week with them?

Jonathan Drummey: The overall remit of the project was to build a set of visualizations and the associated guidance documentation to demonstrate “the art of the possible” for public health data from developing nations with Tableau. We had three goals:

1. Build alternatives to existing visualizations in order to answer the basic questions asked by PATH projects as well as additional questions that are not or cannot be answered by the currently available visualization tools.
2. Create examples to demonstrate how interactivity provides faster/richer insights to help users who have only used MS Excel to become familiar with the cycle of visual analysis and what is possible.
3. Provide guidance documentation of the design decisions and major techniques used in building views including recorded demonstrations.

The background is that PATH is bringing Tableau into more projects in developing nations where the existing data visualization tools are limited to MS Excel and other application-specific visualizations such as those built into DHIS2, ONA, and Open Data Kit. Given those tools we’ve found that we need to do education around what good data viz practices are along what is possible with a best-of-breed tool like Tableau. The generally available Tableau views and dashboards aren’t sufficiently relevant for our country partners, therefore we engaged with The Data School to build out a series of examples that we can use in capacity building around data visualization.

The discussions between PATH and The Data School started informally at one of Tableau Conferences. We found The Data School to be extremely receptive to our ideas of what we needed, honestly the hardest part of this whole project was getting the appropriate clearances on our end for sharing the (blinded) data.

AK: Overall, how does a Data School project work?

JD: We (PATH) brainstormed around several different project ideas and ultimately submitted one to Andy and David at The Data School. Once we collectively had a sense of the scope the next steps were to get agreements in place, ensure the data & supporting documentation was ready, and write descriptions of the use cases that we wanted analyzed.

There was a kickoff meeting on the first day of the project week to go over the use cases with The Data School cohort and they got to work. We had Convo available for quick Q&A threads to clarify details and a mid-week check-in to more formally review progress and offer feedback. Then at the end of the week The Data School gave a series of presentations of results and uploaded all the workbooks and documentation they’d prepared.

AK: What were your expectations going into the project?

JD: We were hoping to see some combination of “typical” Tableau visualizations for our public health data along with some novel solutions along with documentation so new users in our trainings could follow along to build those views using their own data.

AK: Did the team meet your expectations?

JD: Pretty much! We got a ton of workbooks and documentation, some really interesting takes on the four sample data sets we’d provided, and a novel chart type comparing actuals to goal that neither Andy Kriebel nor I nor Tableau’s Jock MacKinlay had ever seen before. (Not to say it hadn’t been created, just that none of us could remember ever seeing it). I won’t spoil the surprise, I’m looking forward to a Data School blog post on that chart type!

My only regret was that the way the scheduling worked out we did the project on a holiday week so The Data School cohort’s availability was abbreviated and that led to some visualizations not having the level of fit and finish that everyone involved had hoped for. That leads to the next question…

AK: Will you be able to put the output from the project into action at PATH?

JD: Yes! Like I mentioned above we need to do some finishing touches on some visualizations before putting them to use in our trainings. However we’re already using one set of visualizations around child vaccination in a training in Mozambique this week, and next month plan to put into production a visualization based on one of the use cases around studying variation reporting rates for malaria incidence in Zambia for the Visualize No Malaria project http://visualizenomalaria.org.

I’d also like to call out an exploratory visualization that was built on an ad hoc basis by one of the Data Schoolers on the sample child vaccination data set we’d provided. It raised some very fascinating questions around gender equity for vaccinations. This is data that had not been reviewed using Tableau before and for me this highlights Tableau’s ability to rapidly iterate across a variety of variables in the data and reveal previously unseen patterns. We’re going back to the actual data to see if those initial findings hold up…if so it would lead to some really interesting work!

AK: What recommendations do you have for others that will be working with The Data School on client projects?

JD: I’ve got three recommendations:

1. Take some time to document your use cases, particularly around the business questions you want answered and the intended audience. And be careful not to make assumptions about what The Data School knows about the use cases. For example, in our situation the largest part of documentation we provided was around terminology and the kinds of users that we are trying to serve.
2. If your use cases are for The Data School to build views and dashboards then also make an effort to clean up the data, document known issues in the data, create a data dictionary, identify join keys, and even build out Tableau data sources and calculated fields. We did a lot to make the data ready and we got some appreciations from the cohort on how good our documentation was, and I think if we’d gone further it would have given them more time to build and less spent on tasks like validating joins.
3. Make sure that there’s room for creativity. One way we did this is that our use cases did not specify chart types, instead we specified business questions that we wanted answered. We saw The Data School as giving us dozens of hours of free labor from some really bright people with great teachers and we wanted to ensure they could have some fun with it and could go in directions that we didn’t expect or imagine. That’s exactly what happened for us, and we’re really happy with the results.

AK: One last question. What’s your top advice for people that are interested in applying to The Data School?

JD: Go for it! Personally I’m a little envious of the experience that The Data Schoolers get and the skills they get to practice and build. Beyond that I suggest experimenting and playing with data sets you are interested in and start building your “muscles” for question-asking and creativity.

AK: Thank you Jonathan for giving us the chance to work with you and with PATH. These experiences are invaluable to us.