Industry executives and experts share their predictions for 2019. Read them in this 11th annual VMblog.com series exclusive.
Contributed by Joy King, Vice President of Product Management for Vertica at Micro Focus
Machine Learning Projects will move from Data Science Projects and Innovation Labs to Full Production Led by Industry Disruptors in 2019
Artificial intelligence, cognitive computing, deep
neural networks, augmented intelligence, machine learning, and so many more promising
technologies have been in our world for decades. But something has finally changed and it's something
that is very important. "Smart compute"
now has the scale and scope of data it needs to learn, something that has not
been available even though the computing technology and predictive algorithms
have been. Storage costs have dropped
dramatically making it economically possible to store the vast amount of data
available to us. To move from historical
reporting to predictive and pre-emptive analytics requires a volume and variety
of data to "teach and train" computers at a level of scale that has never been
available to us... until now. But this
data explosion does bring its own set of challenges.
Virtually every company in every industry has machine
learning (ML) projects, often in Innovation Labs or Data Science
departments. Most of these projects are
reliant on specialty platforms that offer strong capabilities but that cannot
access all of the data relevant to business objectives. All of this data
is stored in a variety of data warehouses and data lakes, none of which have
the ability to run end-to-end machine learning, forcing data movement to the
specialty platforms. Data movement of terabytes, and even petabytes, takes a
lot of time and money, plus it introduces a truly dangerous security risk,
especially if key data contains personally identifiable information (PII). In addition, speciality platforms have
limitations on the volume of data that they can handle. As a result, only a subset of data,
frequently known as down sampling, is used to train and score ML models, resulting
in limited accuracy.
In 2019, current industry disruptors and smart
traditional companies will bring machine learning to all their data, rather
than moving their data to the ML platforms. They will leverage all of the
data available to them - regardless of where that data is stored - and benefit
from the security and governance that protects that data. They will leverage advanced analytics
platforms that integrate and communicate with ML languages like R and Python
that are chosen for model training and evaluation.
These data determined leaders will not compromise on
accuracy and they will not accept delays. Accurate predictions are useless if they come
too late to take any meaningful action. These forward thinking companies will take
the lead in a wide variety of business use cases, including predictive
maintenance on medical devices, predictive revenue based on personalized
customer behavior analytics, and proactive fraud detection and prevention.
And most importantly, they won't settle for knowing
what is going to happen - they will use all of their data combined with end-to-end
ML functionality to produce recommendations and to automate the actions
necessary to influence the outcome.
##
About the Author
Joy King is the Vice
President of Product Management for Vertica at Micro Focus, where she and her
team identify key market trends, customer input, and innovative ideas and
we collaborate with R&D to build the best analytics database on the market.
With more than 25 years in the industry, Joy has experience spanning direct
sales, global account management, sales leadership, industry marketing, partner
management, human resources, employee communications & engagement, product
marketing and product management.