Page 83 - The Real Work Of Data Science Turning Data Into Information, Better Decisions, And Stronger Organizations by Ron S. Kenett, Thomas C. Redman (z-lib.org)_Neat

74 The Real Work of Data Science

He will assume leadership in statistical methodology throughout the company. He will have
authority from top management to be a participant in any activity that in his judgment is worth
his pursuit. He will be a regular participant in any major meeting of the president and staff”
(Deming 1986). While today’s CAO must rely on personal gravitas more than formal authority,
Deming was surely on the right track.
At a minimum, data science should be aligned with a company’s most important strategic
priorities. For example, Carlo Torniai, head of data science and analytics at Pirelli, the tire
manufacturer, concentrates on three main areas: smart manufacturing, cybertechnologies, and
the extended value chain, from the supply of raw materials to the final point of sale. His team’s
brief includes measuring and managing the data more precisely and using real‐time information
to develop more efficient solutions in these domains (Pirelli 2016).
For Pirelli, the biggest source of data is the production line. It measures the operational
parameters associated with tire making and the quality of the product throughout. For any tire,
Pirelli monitors the raw materials used as inputs and the different settings and readings on the
machines that produced it. Armed with this information, Pirelli builds predictive models for
the expected quality of that tire.
The next step is to move from predictive to prescriptive models and adjust machine settings
in real time. The system will “learn” each time it makes a change and, as a result, the process
will be continually improved.
Pirelli’s data team aims to add technologies that predict when tire maintenance is needed,
letting drivers know when their tires should be inflated, repaired, or changed. This allows fleet
managers to keep downtime to a minimum.
“In the not‐so‐distant future we envision a virtual factory where at any given time the allocation
of resources and expected outcome is known, and where machines can automatically regulate
processes and material flow and suggest skills required on the floor,” says Torniai (Pirelli 2016).
Torniai feels that part of his work is to explain this new approach while proving the business
case. “It doesn’t just require technical skills but communication skills and the ability to tell
stories with data to people who are not necessarily technical folks,” he says. “Then it’s about
explaining that often you don’t get a black and white solution but a range of possibilities. So
you need to explain the ‘fuzziness’ in results to people who are used to dealing with straight
numbers, then use this as the basis to make decisions.”
Torniai could not be successful without the full confidence of senior management.

Building a Network of Data Scientists
Further, as we have previously noted, the need for data science in every sector, in every
company, and in every department therein is becoming increasingly clear. Companies, smart
ones anyway, are just beginning to realize that a fundamental transformation, driven by data,
increased computing power, and burgeoning AI capabilities, is afoot. The most important part
of the CAO’s job is ensuring that the company has a network to well‐placed data scientists to
support, and in some cases lead, this transformation.
The worst mistake a company can make is to hire a cadre of smart data scientists, perhaps
organized into a data science lab, provide them with access to the data, and turn them loose,
expecting them to come up with something brilliant (Redman 2018a). Lacking focus and
support, most fail.