Industry executives and experts share their predictions for 2022. Read them in this 14th annual VMblog.com series exclusive.
Trends in Data Analytics
By Luke Han, CEO and co-founder of Kyligence
Kyligence CEO and co-founder Luke Han, has provided his predictions for the important trends in data analytics for 2022. The theme of his predictions revolves around automation and empowering more individuals to do more with data with less effort and fewer impediments.
Metric Stores to Deliver Pure Truth
Even as organizations are enjoying a continuous jackpot of new data to analyze, they wrestle with the challenge of identifying and exploiting their most valuable data. Data teams in 2022 will seek to get a handle on their most valuable data assets. In years past, master data management was the trend that identified customers, products, suppliers, etc. as valuable enough to get extraordinary attention. In 2022, it will be metrics (KPIs).
It has long been understood in business intelligence circles that key performance indicators (KPIs) are among the most valuable. These KPIs are so essential for optimizing business operations, sales performance, marketing effectiveness, product success, and myriad other business processes that they now proliferate in every corner of the digital enterprise.
KPIs by their nature define and track progress of people, products, processes, and business objectives. While they are the daily fuel for departmental managers to track the progress of their teams and initiatives, general managers, division heads and C-level executives are seeing value in aggregating, standardizing and evolving KPIs en masse to get better visibility into how their organizations are performing. And they want to see this with more detail than the aggregated and interpreted reading of those metrics through traditional reporting processes.
With this proliferation of metrics, data teams and savvy executives are now seeing the collective value of maintaining a dedicated metrics store that creates a specialized single source of truth for KPIs. Without a metrics store, companies risk duplication, inconsistency, contradictions, and inaccuracy in their most important and valuable data assets.
Data as a Service and Data APIs
Platform as a Service (PaaS) API sets abound in all of the popular cloud platforms and are an integral part of the growing API economy. Data APIs are beginning to follow suit and will gain momentum in 2022. While SQL/NoSQL remain very popular access methods for data, many see the use of REST APIs to access data as a better match for microservice architectures which are increasingly becoming the default.
As people pay more attention to their own API economy, the ability for REST APIs to to handle data resources more efficiently than SQL may further accelerate their adoption. This is also the reason why Data as a Service will become a more common design goal for data teams as well as ML teams that need a steady stream of high quality data delivered just in time. These teams may look to build their Data as a Service on open source platform APIs such as Apache Kylin or PostgreSQL.
Data Consumer be Augmented by AI
With more and more organizations looking to exploit machine learning and AI in their operations, 2022 will be a year when the larger audience of data driven professionals – citizen data scientists or perhaps more accurately citizen data analysts – will begin to see practical benefits of machine learning in their daily professional lives. New ML development frameworks (PyTorch, TensorFlow, and H20) and MLOps platforms (Algorithmia, Comet, Kubeflow) will enable more data consumers to work on machine learning projects on their laptops and then easily make them operational in the cloud at scale.
With the increased accessibility, we will see ML projects lurch forward faster than we would have thought possible a year ago. But the tools and the frameworks themselves are beginning to get ML-powered makeovers with offerings like PyTorch Lightning and TensorFlow/Keras. Both of these frameworks are becoming wildly popular and are maturing fast.
In fact, that process of maturation will be measured how the ML frameworks can themselves benefit from the machine learning techniques that they enable. While the concept seems a bit like a serpent devouring its own tale. It is a key step forward in helping data consumers transparently apply and benefit from the progressive intelligence possible from AI.
Today, it takes far too long to become an actual data scientist for the world to mint the millions needed to realize the levels of intelligence and automation that we all envision. To truly break through to our notion of the sentient enterprise, an unprecedented leap forward in the use of data requires that every information worker should be able to expect some help from the systems they use. In the past, it was to help functions and training. Increasingly it must be AI, and that AI must seem natural or be invisible. AI will increasingly empower those who lack the technical skills of a “power user”, by creating a bionic analyst. To achieve this, software vendors will need to improve their understanding of how data consumers use and gain advantage from data.
##
ABOUT THE AUTHOR
Luke Han is Co-Founder and CEO at Kyligence, co-creator and PMC chair of Apache Kylin project; In past few years he had been working on growing Apache Kylin’s community, building ecosystem, and extending adoptions. Prior to Kyligence, he was the Big Data Product Lead at eBay. Prior to eBay, Luke was chief consultant at Actuate China.
