TECH | 16 Feb

A Data Science Lab for enabling Digital Innovation

How and why a Data Science Lab is created: the Engineering experience

“Our company seeks simple answers to complex problems. This is often not possible or wrong.
But there is always a simple way to find an answer to a complex problem.”
(extract from a public speech, June 2003)

Manager who need to innovate their solutions knows that they must have three aptitudes: the habit to “look around” with curiosity and interest; the capacity for synthesis and choice; the aptitude of not “reinventing the wheel” but of enriching solutions, also using what is already present on the market, adding their “secret sauce” which adds that unique flavor for creating originality and distinction.

If they lead team of innovators, they must know how to create a multidisciplinary and multicultural environment in which these can grow over time both as individuals and as groups. It is a question of enhancing the extraordinary aspects that brilliant engineers possess, such as creativity, talent and passion, but also of managing relational complications and psychological implications, with the support of operational instruments and practices in a community setting that enables the harmonious growth of a common culture.

The main activity is to create a team that is pre-adapted to meet the needs of innovation when the conditions are appropriate.

It comes down to managing a flat and agile organization that promotes the exchange of ideas, information and knowledge, bringing out those potentials able to anticipate the future and not simply to pursue it: an organizational context generating a pool of potential made up of resources which, through skills and relations, is able to respond to different events when these occur. Knowing well that dealing with exploratory activities could require a “U-turn” depending on the “way the wind blows” and that the sustainability of the whole is demonstrated by the patient, but progressive, collection of tangible results, even if of a different nature.

This approach gives rise to five lessons:

  1. If it is natural to invest in people in information technologies, which are high intellectual labor-intensive activities, this is essential in innovation: it is a question of creating the incentives for working groups to emerge and manifest features such as talent, creativity, discipline, as well as an appropriate rate of productivity, in a balanced manner.
  2. A flat organization is needed to foster the right conditions: it is about creating a group of professionals where singularity can emerge, but especially where the result and the growth of the individual are the result and the growth of the collectivity.
  3. Managerial skill lies in the ability to have a view of the whole, in being able to manage complexity using multi-disciplinary approaches that are essential in an economy based on services and knowledge.
  4. Having information circulate and sharing knowledge are fundamental practices for obtaining often unexpected results: the experience of recent years has shown that terms like openness, mutual aid, gratuitousness and trust are not words from a “book of dreams” but can be everyday practices when inserted in the right context.
  5. It is necessary, if sometimes exhausting, to keep an open mind for living the present, investigating novelties, knowing what is emerging or is on the peak of innovation and – where possible – anticipating the future.

To create a community of work, it is not however sufficient to organize the context: it is necessary to have a common goal and this sometimes given by events “that happen”.

Engineering’s Data Science Lab

In 2005, Peter Sondergaard, head of the Gartner research firm, stated in a public speech: “Algorithmic business is here“. And he continued by saying that “interconnections, relationships and algorithms are defining the new business“, which will be “driven not only by data, but by algorithms“.

Having listened to these phrases and aware of how market trends were evolving, we decided to prepare ourselves by building our “pool of potential”. We had knowledge of technologies and instruments and keeping up to date on what was emerging was not a problem at all; we were building what is our private cloud infrastructure that we use today, not only for in-house developments, but also as an offering on the market. But what was missing in a structured fashion was new specialist skills. At this point, both in terms of in-house collection of talents and skills that operated in isolation and of selection from the market of resources with experience and bright recent graduates, we have set up a multidisciplinary team with a common denominator: passion for the analysis of the data in order to extract value. Today, at the Big Data and Analytics Competence Center, a team of physicists, mathematicians, statisticians, engineers, economists, chemists and computational linguists work alongside “traditional” big data architects and developers.

There was thus the need for a common goal: building a platform which included agile and outcome-driven methodological approaches, skills, techniques and instruments adapted to support digital business. This was addressed with a change in approach: not only starting from customer requests (from knowledge of their business model to analysis of processes for responding to a business opportunity) but, on the flip side, by using data and algorithms in exploratory way to anticipate demands, observing what happens (from perception of a business opportunity and identification of appropriate processes up to achieving adaptation of the business model).

Engineering’s Data Science Lab is an infrastructure that uses state of the art technologies and instruments (Hadoop solutions, of data analysis and of front-end) where statistical techniques are used, and machine learning, and artifacts (reusable and combinable analytical components) are consolidated, both for in-house research activities and for the offer of services to customers. Several cases of use and practical examples applicable in every market have already been addressed: predictive analysis of customer loyalty and turnover forecasts, predictive maintenance of plant production lines, automatic evaluation of the delivery and effectiveness of loans granted, analysis of signals from devices connected to the power supply of housing, social listening with joint analysis of texts and images, and functionality of the typical recommendations of e-commerce.

Today, advanced analyses based on combined techniques and algorithms, some also original, can be realized relatively quickly to support the entire Analytics Maturity Model (Descriptive Analytics, Diagnostic Analytics, Predictive Analytics, Prescriptive Analytics).
Rather than using algorithms developed, the result of this experience, which is still in progress, and is already being used in various fields of realization, is given by the experience itself. An approach to innovation aimed at creating a core of expertise capable of growing not only as the spread of a new way of understanding business and of processing data that today give it a new form, but also of being at the disposal of partnership initiatives with customers to accelerate the development of their internal skills.
Without neglecting cooperation with those “ecosystems of knowledge” which we have always frequented as Engineering thanks to a solid experience both within the open source community and as part of Italian and European research initiatives and projects.

Agility, capacity for exploration, openness and sharing of knowledge are the skills that a company must have to be a player at the frontier of digital innovation.


Gabriele Ruffatti