Scientific Games Lead Data Scientist Haley Bobo borrows from the Harry Potter universe to describe data science as “somewhere between Arithmancy and Divination.”
Since joining Scientific Games in July 2018, she has been immersed in the world of lottery statistics. But she doesn’t have a crystal ball. “In simple terms, we use data to help answer tough questions. The answers aren’t necessarily apparent when looking at a bunch of numbers. Statistics, model building, coding and research all help us understand what will happen, what lotteries can do to responsibly drive profits, and what gaps we see in their portfolio—all from a mathematical and scientific perspective,” said Bobo.
Bobo stressed the importance of understanding the difference between general analytics and data science. “Both are rooted in the same thing—historical data. From an analyst’s perspective, they are really trying to study the past data to find patterns, trends, and anomalies. Now as data scientists, we take this one step further. We use those patterns and trends to help us forecast. We statistically measure those anomalies to try and find the root cause. My favorite way to delineate between the two is an analyst typically answers the question ‘what happened when?’ whereas the data scientist answers the question ‘what would happen if?’”
Bobo and her team must analyze, process and model data then interpret the results to create actionable plans for Scientific Games’ clients. Bobo said, “99% of the requests that we work on come from the lottery or their account management team.”
Data science research is geared to custom solutions for lotteries. Each lottery is different. “They all need special attention. And the best thing is that our Scientific Games Enhanced Partnership (SGEP) analytics package offers a ton of flexibility. While we typically hire analysts at our SGEP sites, they do hear about other data science related problems. In these cases, we collaborate to reach a solution. What’s critical is that the analysts get to work with the lottery every day and understand their business intimately. It is so interesting to me because they have unique perspectives,” said Bobo.
While some of the data could seem unwieldly to fit into the various databases the company utilizes, Bobo said it just requires some thought about the data manipulation. “We use a lot of weekly sales data as well as game data that gets sent to us from each jurisdiction. But it comes in a variety of formats. We have excellent data engineers who work hard to process data and put it into the different platforms we use in a uniform way,” she said.
COVID-19 World
Even with “data divination,” no data scientist could have predicted the impact of the global pandemic declared in 2020. COVID-19 has proven that data science has its limitations.
A key limitation during the pandemic is uncertainty, which leads to unsteady predictions. “We’re operating in an era where everything is uncertain. All because of another key limitation: lack of history. Data science at its core is built on historical data. We are now in a situation where we have no historical data that represents a global pandemic,” said Bobo.
“Every day we are learning new ways to adapt to the lack of historical data. We are not unique. Every industry is struggling. I don’t think anyone foresaw the lottery industry growing as it has during the pandemic. I think the question now is more about sustainable growth,” said Bobo.
Breakthrough Tool
One of the projects that Bobo is most excited about centers on prize structure variety. Last year, Scientific Games had a lottery request an optimum prize structure analysis. “When we did our due diligence, we found that within a given price point, the lottery was using the same prize structures over and over again,” said Bobo. “We thought there has to be a way to quantify the variety that a lottery has at each price point. Through this process we developed something that we are calling a variation score. It is measured at a game level.”
This tool looks at the unique prizes offered by each game as well as the number of winners and the allocation of prize fund. “This gives us a solid number to say this game is either like a ton of other games or unlike a ton of games in a lottery’s portfolio. Then we can roll those up together to create an overall variation score by price point. There is a ton of potential here,” said Bobo.
Scientific Games is now using this tool with another lottery to find out where they can best optimize their prize structures. “Test and learn is very difficult with our type of product. To have a lottery willing to put in the effort and see what works with their players is music to my ears,” said Bobo.
Background
Before earning her master’s degree in the School of Applied Statistics at Kennesaw State University, Bobo spent five years teaching mathematics. This experience comes in handy when explaining the data and findings to her internal and external customers.
She is not surprised that more women are attracted to a career as a data scientist. “This job perfectly blends art and science. It requires a bit of multi-tasking, storytelling, and statistics—all things I believe women are inherently good at.” said Bobo.
Bobo placed second in the ATLytiCS Data for Hope competition at the 2019 Southern Data Science Conference in Atlanta. “My co-worker, Malu Gopi Punnackal, teamed up with me. We were handed a data set on human trafficking and asked, ‘what can you do?’ The competition truly gave us the chance to do some good in the world.”