Background
Over the past month, I had the opportunity to design the analytics problems for the first-ever Alteryx E-Grand Prix event (Figure 1). The Alter.ECO event sponsors were kind enough to ask me to create the problems used during the competition. This four-round Alteryx internal competition was intended to help Alteryx staff learn new skills and techniques while also learning about an ecological issue we are all familiar with.
The topic chosen for the competition was global climate change. Luckily, I had gigabytes of data lying around that was perfect for such a competition. The main challenge for me was to walk a fine line between chaos and creation, as you will see.
The Goals of the Competition
As Sir Paul McCartney sings in Fine Line (2005):
There is a fine line
Between recklessness and courage
It’s about time
You understood which road to take
It’s a fine line
And your decision makes a difference
Get it wrong you’ll be making a big mistake
Paul McCartney (Chaos and Creation in the Backyard, 2005)
What this means to me is that there is a fine line between designing a fun and exciting analytics competition or one that fails because competitors could not solve the problems in the time given.
The reason that I included these lyrics is two-fold. First, I had the pleasure of attending a Sir Paul concert a couple of months ago, and it was a delight (slideshow at the end of this article). The man is a total musical genius and is a stellar performer (even at 80 years old!). Second, designing an analytics competition makes you carefully distinguish between problem simplicity and an appropriate degree of complexity.
Your primary goal is to create a series of challenging problems that advance from simple to complex through four competitive rounds. While doing this, you should strive to weave a data story that educates the contestants about the competition topic as they move through the rounds.
By the time you reach the final-four contestants in round 4, you better be sure to have crafted a problem that can separate the finals challengers, that provides for competitive drama, and is one in which the solution can be obtained within the specified time limit. What this means is that there are multiple competing objectives, which makes you walk the fine line between chaos and creation.
You must produce a creation that makes sense, is well-documented, and can be understood within the timeframe of interest. If you fail to do this, chaos will ensue, with no winner being crowned. For all the work that goes into the planning of an event like this, having chaos erupting in the final round was my greatest fear.
The Keys to Creating A Great Competition
Designing the computational challenges (i.e., problems to be solved) is only a small part of creating a great competition. More important is having a strong team dedicated to promoting, documenting, scheduling, and getting the online platform set up for the finals.
This team will also be responsible for grading rounds 1-3, and contacting the contestants to tell them whether they advanced through each round. There are bookkeeping and problem grading activities that are a part of this task. There are rules to be written and methods designed for effectively documenting everyone’s independent work. If the team does not execute these tasks flawlessly, there will be teeth-gnashing and unhappy contestants in the competition.
Another important job for the team is to recruit independent test takers that are bound to secrecy. These are the first people exposed to the problems. If they cannot solve the problems in a reasonable timeframe given the instructions, the problems either need to be changed or better documented. The communication iterations between the problem designer (me) and the test takers (awesome Alteryx staff and the incomparable Alteryx Ace Claire McCollough) were very important in building the competition problems.
Designing The Competition Schedule
The Alteryx team designed the schedule and planned the activities for this event, which spanned 12 days in August 2022. Figure 2 shows the final schedule for this inaugural event. Getting the schedule right was a critical component of the job the Alteryx team had to do.
Designing the Competition Problems
The competition problems need to be designed with several goals in mind. Before the competition could even begin, however, the contestants had to be given specific information in two preliminary activities to help them prepare for the upcoming challenges.
Preliminary Activities
- Based on a list of Alteryx skills (concepts) needed to compete in Rounds 1 – 4 (provided by me), the Alteryx expert Nicole Johnson designed a generic workflow to teach the skills and concepts. Watching Nicole explain various solution techniques reinforced for me how superb she is in working with Alteryx Designer. It seemed like I was watching Banksy create another invaluable piece of street art.
- I completed a live lecture that discussed my global warming research, the insights gained, and the data used in the competition. I also dropped some big hints that would be needed during rounds 1-4 of the competition. I hoped these hints would help keep a couple of hundred people awake during my 30-minute presentation! I also knew that this was my time to teach, to explain what I’ve learned by studying global warming. Once the games began, I predicted that people would turn off their “learn mode” and focus on their “problem-solving abilities”.
Competitive Rounds
Although there was no set time limit for rounds 1 to 3, we chose a 20 to 30-minute range to achieve the solution. This time range helps determine the complexity of the problems. The experience of the test takers ensured that we built solvable challenges within this time window.
Round 1
This challenge should be solvable by nearly everyone in the competition within a 20 to 30-minute window. This problem should be a warmup to the data, with limited scope and relatively easy solutions. Concepts needed to achieve the solution included:
- 1. Unioning
- 2. Assigning and/or changing data types (crucial skill for the date field)
- 3. Creating a two-part custom Filter
- 4. Extracting date components (creating the month)
- 5. Aggregation by groups
- 6. Strategic sorting
- 7. Ranking results (stretch goal)
- 8. Filtering to select the correct answer
Round 2
This challenge should increase the problem complexity and include more advanced topics. This round should begin the weeding process, removing contestants that are not yet ready to advance towards the finals. Concepts needed to achieve the solution included:
- 0. The ability to carefully follow instructions!
- 1. Computational awareness
- 2. Joins by record position
- 3. Unit conversions
- 4. Percentage calculations
- 5. Computing correlation coefficients
Round 3
This challenge increases the problem complexity and is the round in which the final 4 contestants are selected. Although many people may solve the problem, it is the people with the four fastest solutions that advance to the finals. Concepts needed to achieve the solution included:
- 0. The ability to carefully follow instructions!
- 1. Working with two different types of data sets
- 2. Creating a two-part custom Filter
- 3. Creating spatial objects and computing distances between points
- 4. Working with location pairs (i.e., a trip beginning and a trip end)
- 5. Performing multiple joins leading into a union
- 6. Computing across multiple levels of detail to find the correct answer
- 7. Merging pieces of data from various threads in a workflow
Round 4 – The Championship Round
This is the problem having the highest challenge. In this problem, there are four solutions. Four solutions allow the competitors to reach 25, 50, 75, and 100% completion percentages. This allows spectators to gauge who is in the lead or lagging.
In this event, the first two problems were essentially easy, but the gauntlet got dropped on the third part of the problem. This is the point where I expected to see a separation between the competitors, with people strong in predictive analytics able to complete the solutions while others struggled. That expectation came to fruition. Concepts needed to achieve the solution included:
- 0. The ability to carefully follow instructions!
- 1. Understanding what p-values represent
- 2. Data blending concepts, aggregations, seasonality
- 3. Being able to interpret and use results from regression models
In the Round 4 finals, the competition was fierce with multiple people seemingly in the lead at various times. This means we achieved drama in the finals. As hoped and expected, the third question was the separator and the proved to be a question that walked the fine line between chaos and creation. It was a great competition, and Figure 3 shows the electronic Kumospace used during the competition.
Final Thoughts
An hour after writing this article, I took my son to his guitar lessons. When we pulled into the parking lot, I looked up to see this sign (Figure 4). Was it a sign that I was on the right track when writing this article? I think so!
Thank Yous
I offer many thanks to the Alteryx Team and others that helped me during this project. Although it is likely that I’m going to forget to add someone to this list, I’ll always remember your contributions! Thank you very much:
Jenny Yuen – for recognizing the possibility of how my work could be used in this way, and for being the driving force and organizer of this event (Alteryx give her a bonus, please!)
Margaux Sellnau – for handling all the details with such professionalism and wisdom (Alteryx, please hire her when she is done with school!)
Savior Willams-Onuorah – for being really cool, asking great questions, and introducing us to the Kumospace!
Joshua Burkow – for being as cool as Paul Newman, Robert Redford, Clint Eastwood, and Matthew McConaughey all combined and running the show like a total boss!
Nicole Johnson – for running a training session with such command, it was a superb performance!
Claire McCollough – for helping complete the final workflow with a critical eye towards solvability and documentation, as well as your expert commentary during the finals. Your future is going to be amazing!
The others – Jimmy Garrett, Cailin Swingle, Judson Woods, Jessie Chou, and other contributors I might not know about. Without you guys, we would have had chaos erupting!
Thanks for reading!
