Getty Images/iStockphoto

AI, graph databases among top BI and analytics trends

Leading trends in BI and analytics include machine learning, multi-cloud and graph databases. IT expert William McKnight details in a webinar how they'll affect enterprises.

Machine learning, graph databases and multi-cloud -- William McKnight, president of McKnight Consulting Group, sees these as some of today's top enterprise BI and analytics trends.

McKnight advised IT professionals to seize on important BI and analytics trends and work on putting the technolgies to use in their organization to help build a more advanced analytics future.

"What we really need are champions and leaders out there that are going to pick up the helm and find ways to get the stuff in the enterprise and get it moving towards bottom-line business success," McKnight said in a recent Dataversity webinar. "Momentum is paramount."

McKnight culled his trend predictions from several large-scale surveys he conducted, as well as from observing his clients. Although some of the BI and analytics trends that he detailed may not surprise forward-thinking IT professionals, they're all already starting to affect large enterprises in tangible ways.

From BI to AI

William McKnightWilliam McKnight

In the webinar, McKnight said the best use for enterprise's data will be training AI algorithms.

"AI algorithms are going to be important, and they are obviously trained on great data," he said. "So, a mantra of mine is: 'Let no data escape.' If you've been letting data escape, you are just hamstringing your future AI efforts."

In an interview, Forrester analyst Boris Evelson said he agreed with McKnight's view, adding that AI is unquestionably one of the biggest BI and analytics trends in 2019 -- specifically, machine learning. "With machine learning, BI professionals can … find and identify every kind of pattern, trend, correlation, signal," Evelson said.

McKnight said his clients often ask him how to architect the data management environments for what he sees as the inevitable AI-dominated future. He said his best answer is to create a strong data infrastructure. This includes effective foundational structures, such as a data lake, data warehouse or data mart.

A mantra of mine is: 'Let no data escape.' If you've been letting data escape, you are just hamstringing your future AI efforts.
William McKnightPresident, McKnight Consulting Group

McKnight said he's also challenging his clients to change any initiative they've labeled as BI to AI and think about doing that initiative as AI.

"If you think of the initiative as writing reports, building key performance indicators or something of that nature, you're never going to get there in terms of thinking about it as AI," McKnight said. "You have to get behind the reasoning for the report and the reasoning for the KPI."

"You have to understand why something is being done, and oftentimes, when you get to that why, that is what you can disrupt with artificial intelligence," he added.

That doesn't necessarily mean AI will dominate the BI and analytics trends in coming years, Evelson said. "A more appropriate statement would be enhancing or improving, as in 'improve BI/analytics with AI,'" he said.

A good way to introduce AI into enterprises is through automation, McKnight added. He said he's seeing many BI vendors using AI to automate processes and insights.

'Year of the graph'

McKnight declared 2019 "the year of the graph," saying that IT professionals should take advantage of the many graph capabilities that already reside in databases. Organizations will find graph databases -- a type of NoSQL database that uses graphs to store, map and query data relationships -- more and more useful this year and in the years ahead, McKnight predicted.

"It's not just for the nice, pretty interface that it provides us with all of the nodes where we can drill in and we can see who's connecting and whatnot," McKnight said. "What I like about [a graph database] is the speed with which you can find connections and the relative importance that you can find out about the nodes on the network in it."

Graph databases "really help you understand what the important things are for you to focus on in an enterprise," he added.

Evelson agreed with McKnight but went further: "I'd say the trend is about having the right tool for the right job. All sorts of NoSQL -- including graph database management systems -- will continue to become increasingly popular."

Data visualization footprint expands

Another one of the top BI and analytics trends for the coming year, according to McKnight, is progress in traditional data visualization methodologies.

"Every enterprise out there is embracing data visualization to some degree, but I think it's going to really take off [this year] because so many organizations are still doing things the legacy way and reporting," McKnight said.

He clarified that conventional reporting can still be valuable and is needed in some cases. But any time his clients are thinking in a reporting paradigm, McKnight said he encourages them to think in terms of data visualization instead.

Visualizations don't have to be advanced, McKnight added. He said an organization can enlarge its data visualization footprint by using techniques such as sparklines, bullet graphs, scatter plots, treemaps and infographics.

Evelson, on the other hand, said the data visualization revolution -- going beyond static reporting -- has already occurred in the enterprise.

"There isn't a single BI/analytics platform out there that doesn't include data visualization. In fact, Forrester no longer sees data visualization as a separate market category -- rather, it's a capability of all BI/analytics," he said.

Meanwhile, David Menninger, an analyst at Ventana Research, said in an interview that data visualization has been commoditized, and while incremental improvements in that area will continue, other techniques such as natural language will become the "hot topic" in analytics interfaces.

Multi-cloud becomes the norm

When selecting a data management platform, organizations more and more want to be able to use different cloud configurations to accomplish varied data and business objectives    , McKnight said. That's why the concept of multi-cloud is gaining traction.

"Disparate data-related objectives are difficult to pursue with agility in a single-cloud strategy," McKnight said. "We need multiple clouds -- we need our private cloud, we need multiple public clouds and maybe we need a software-as-a-service provider cloud that provides some unique capabilities."

Next Steps

Graph analytics being used in childhood cancer research

Dig Deeper on Data science and analytics

Data Management
Content Management