DEMOCRATIZED DATA
Why it’s hot: Quality data and democratized data access “are, without a doubt, the backbone of any modern-day company’s ability to be successful,” said Jason Hortsch, data scientist at Getty Images. With the right tools, “we increased our data democratization, allowing for more people than ever before to efficiently access data. It has allowed our data science team’s work to be scaled out to far wider audiences, allowing for that work to be more impactful. For example, at the beginning of COVID, we were able to detect the increase in searches like ‘work from home,’ ‘social distancing,’ and ‘video call’ by technical and non-technical team members, which allowed us to respond more promptly.”
Emerging or widespread? Data democratization platforms are “still being adopted by all corners of the business, but it is the go-to platform for our reporting,” said Hortsch. “More and more decision makers now have access to accurate and updated data in order to have the confidence to make the decisions that are necessary in today’s fast-paced world.”
Gotchas: Getting buy-in. “As with any new technology or tool, there is a certain level of momentum required to attain buy-in from the company as a whole,” said Hortsch. “It’s not as easy as simply switching to a new tool with a snap of the fingers—rather, there is a learning curve.”
5 years from now: “Decision makers at companies of all sizes will be more comfortable and confident using data to make big decisions,” said Hortsch. “This will require a company’s data to be reliable and easily accessible. Business insight solutions will be the frameworks that allow this—extremely powerful, extremely performant solutions that are still easy to use.”
DATA OPERATIONS (DATAOPS)
Why it’s hot: “There has been no single idea that has been as effective as DataOps at improving efficiencies and encouraging creativity in analytics teams,” said Chris Bergh, CEO of DataKitchen. “An automated orchestration capability that creates self-service sandboxes slashes cycle time and decreases the cost of innovation. When cycle time is dramatically improved, an organization can explore dozens of analytics ideas in the time it used to take to implement one.”
Emerging, or widespread? “Experience from previous recessions suggests that C-level executives will defer discretionary spending and focus resources on pragmatic initiatives that directly contribute to top-line growth and bottom-line efficiencies,” said Bergh. “DataOps impacts both, so many DataOps initiatives are being accelerated.”
Gotchas: Simply getting started. “Focus on finding and eliminating the bottlenecks that slow down analytics development,” said Bergh. “This could mean speeding up deployment, continuous deployment, eliminating errors, or even improving the productivity of data scientists working on models.”
In 5 years: “Markets will change unpredictably,” Bergh said. “Executives will want fast responses to their analytics requests. As organizations face pressures, tempers are going to get short. Leaders of data organizations must prepare for this critical phase by slashing cycle time. When data-driven decision making can make all the difference, data engineers, analysts, and scientists are an enterprise’s most precious resource. Organizations need them focused on creating new analytics that will guide the enterprise through turbulent times. These dynamics will shift focus away from the latest shiny tool and toward reducing cost and producing better outcomes in the way that data teams develop, deploy, test, monitor, collaborate, and measure their analytic operations.”
MACHINE LEARNING OPERATIONS (MLOPS)
Why it’s hot: “Predictive analytics and artificial intelligence present tremendous business opportunity,” said Sivan Metzger, managing director, MLOps and governance at DataRobot. “However, the problem is that once data is prepared and models are created, as many as 87% of them don’t make it into production.” And, even when they do, it is unlikely that they would be managed as required to deliver continuous value over time, Metzger said, adding “They need to be seriously thinking about their MLOps strategy.” MLOps is a combination of technology and industry practices that are designed to provide a scalable and governed way to deploy, monitor, and manage machine learning models in governed production environments, he said.
Emerging or widespread? MLOps consists of relatively new and emerging processes and supporting sets of technologies, but is rapidly gaining traction and becoming more and more widely used, said Metzger. “As more companies seek to finally derive value from their investment in machine learning, they’re increasingly adopting MLOps strategies to take them across the chasm and into production.”
Gotchas: The human factor. “There is “lack of alignment between the different stakeholders around roles and responsibilities, and lines of delineation regarding accountability of machine learning deployments,” said Metzger.
In 5 years: The pace of data, predictive analytics, and AI use in the enterprise will only increase as more companies seek to maximize profits, optimize operations, and ride out disruptions with minimal risks and damages, said Metzger. “However, to get these important insights and benefit from them, models will have to be fully deployed and be able to deliver value over time. We expect that as AI is adopted more over the next 5 years, the road leading there will be paved by MLOps.”