CASE STUDY
Big Data
Predictive Analytics
By Oliver Guy, Retail Industry Director, Software AG
Introduction
Oliver Guy explains
why companies
cannot afford to
turn their backs on
predictive analytics
In the era before predictive
maintenance, manufacturers had just
two options concerning their expensive
equipment. They could run it until it
failed, or they could estimate its useful
life and then retire it before it started
making strange noises and ground to
a halt. Similarly, retailers only knew
their inventories were low when they
ran out, or when someone went to the
warehouse and uncovered the true
picture with a stock-check. Financial
institutions lost millions or billions of
dollars due to fraud, which was only
discovered long after the fraudsters had
vanished or changed jobs. Now all these
headaches can be prevented through
the power of predictive analytics. In
the last few years the technology has
outgrown the hype and produced realworld applications that are being taken
up in many different sectors. It’s now
being wholeheartedly adopted in fields
as varied as manufacturing and logistics,
retail and hospitality, financial services
and telecommunications. The goal is
to radically improve efficiencies, reduce
costs, create new revenue opportunities
and take customer satisfaction and
loyalty to new levels. By continuously
monitoring the actual conditions and
actions of equipment, staff, inventories,
trades, and anything else that impacts
a business, data can be analysed and
acted upon. The aggregated amount
of data is mind-boggling, described in
terabytes, petabytes and exabytes.
Key Advances
The key advance in building predictive
analytics has been the use of statistical
models with historical data. It can
now be deployed to foresee where
bottlenecks will occur so that shipping
delays can be prevented, fraud stopped
before it happens and equipment
fixed before it breaks. In retail, store
operators can order more inventory
before it runs out. Predictive analytics
can be used to determine when a
retailer’s competitors are likely to be
lowering prices, prompting automatic
pre-emptive action via digital shelf-edge
labels. Indeed, within a store, sensors
can also automatically signal when
shelves are likely to be low on goods,
alerting staff via smart badges. In the
logistics industry, predictive analytics
allow supply chain managers to receive
a definitive time of arrival for shipping,
based upon a dynamic statistical
prediction model. In manufacturing,
data streaming from single components
or entire pieces of equipment can be
used to predict the possibility of future
failures, allowing the arrival of new
components to be synchronised with
that of the repair technician.
Big Data
The challenge was to find a cost effective solution without
compromising on security.
30 NETCOMMS europe Volume V Issue 5 2015
The key requirement, of course, for
successful deployment of predictive
analytics, is for an enterprise to be
able to analyse fast flows of big data.
These will stream through from its own
operations and from relevant sources
in its customer-base, market or news
channels. The volumes are so big they
cannot be fathomed without the use of
data scientists, computing power and
algorithms. Once mistakenly considered
lonely geeks, data scientists now
have some of the most desirable and
in-demand jobs on the planet. Using
computer and mathematics skills along
with their native curiosity and creativity,
they mine mountains of data to find
competitive opportunities and to predict
likely future outcomes. Possessing a rare
skillset, they are however, in short supply,
which is why firms need to become
more creative, integrating data science
more tightly with IT departments and
building teams that include computer
experts, mathematicians, statisticians
and business specialists. All these talents
are needed if an enterprise is to crack
the big data code and really drive value
from it.
Conclusion
As big data becomes more accessible,
in part through increased adoption of
open data standards, and as analytical
tools become more readily available,
more enterprises will enjoy the benefits
of predictive analytics. We will soon see
extraordinary, game-changing use cases
where goods are automatically ordered
and delivered to the warehouse before
a sales campaign causes a shortage.
Haulage trucks will meets ships as they
arrive at port and deliver their products
on time, with every traffic and weather
delay taken into account. A continuous,
smoothly flowing logistics network will
result from the greater synchronisation
that a predictive capability will permit.
In the financial markets, trading patterns
will set off alarm bells about the threat
of insider trading, allowing a bank to
take action so that regulatory breaches
do not occur, with the concomitant
risk of hefty penalties. Predictive
analytics using data from sensors fitted
to a patient will even give doctors
the ability to call a man with a heart
condition and tell him to get to hospital
immediately because he is going to go
into cardiac arrest tomorrow and they
need to intervene to save his life. This
is predictive analytics. In an uncertain
world, one development we can predict
is that it will deliver huge benefits for
any enterprise that has the foresight to
invest in it.
www.netcommseurope.com