How Data Science Evolved in 2020: Q&A With Jon Shepherd

2020 felt like a rollercoaster at the best of times and a highway with no exits at others. However, it’s no secret (or surprise) that the massive uncertainty created by the year’s events led to a need to understand the landscape better. From COVID-19 to natural disasters to a ferociously fought US election, 2020 was unlike any other year when it came to throwing curveballs at almost every industry and forcing everyone to find new answers. 

This was great news for data science and machine learning (ML), and the industry’s upward trajectory quickly accelerated. We wanted to look back at how the ML industry fared in 2020 and what it’s looking like in the coming year. We sat down with Jon Shepherd, Explorium’s VP of Enterprise Sales, to look back, look ahead, and talk about the biggest surprises of the year. 

In a year full of crazy twists and surprises, what was the biggest news or innovation of 2020? 

Obviously, the pandemic created a big surprise for our economic climate, which forced us to rethink our approach to ML and data science. This, in turn, created a rethinking of the entire approach to ML and data science. Software vendors and the data science community had to determine how to adjust to this new reality quickly. For data scientists, this meant a return to rules-based decision making using whatever data was available. For data and data science platform vendors like Explorium, it meant helping customers find signals in new, fresh data. Who saw this coming? 

How do you think COVID-19 impacted data science knowledge and implementation this year? 

Good question. I feel like COVID exposed the importance of data in data science.  It also caused companies to review what processes can be automated. Combined, I suspect that the economic climate created by COVID will be viewed in hindsight as a launchpad for data science implementations across industries in 2021.

What data science trends did you spot in 2020 that drove the sector the most? 

There were a few, I think. The first was the commoditization of ML models thanks to the growth of open source languages and ML platforms. The number of available options meant that the real distinction between each of them was small, and to get a real uplift, you needed to look elsewhere — in this case, to data. The next trend is related, but I think the maturation of AutoML platforms was a major driver in adoption and expansion. The industry offering real solutions to data science problems meant that more companies could enter the ML game. 

Next, I think another major trend was the evolution of ML and data science into C-level issues that can drive corporate value and even help companies survive this economic climate. More executives understood the importance of data science and ML, and it was a major drive in the expansion of the industry and its mainstream adoption. Finally, I think the industry had to react quickly to the expansion of regulatory policies worldwide, such as GDPR and FCRA. Organizations are now more hard-pressed to access the data they need. 

What are some of the biggest trends you see heading into 2021? 

First, I think how the US economy recovers after 2020 will be important, especially whether technology investments drive it. 2020 was difficult, but data science and tech were pretty reliable as organizations scrambled to adapt and implement new solutions that could prop them up. The second big trend I’ll be looking out for is whether the major platform vendors — Amazon Web Services, Google Cloud, and Azure — will realize the massive growth in the data sector. If they do, could it lead them to start incorporating data products and platforms directly into their data science offerings? 

Which industries do you think will dive in the deep end and embrace ML and data science most actively next year? 

Financial services have been leading the way in ML adoption for several years now, and I think 2021 will see that trend continue as well. However, I might approach this question differently. I think most verticals should be using data science platforms like Explorium but might avoid it because they operate in low-margin industries, and they can’t compete for data science talent. I think the data science field needs to find ways to make their platforms more accessible to these organizations. 

Lastly, anything you’d love to see happen to data science and ML in 2021? 

I would like an analyst to declare that there is little innovation happening in AutoML, ML workflow, and model management, and the next big thing is data automation!