Data science, which processes and analyses large volumes of data, is a booming activity. It is taking its place in resource management and is shaping a new expert profile: the data scientist.
It is now a well-known fact that, never before, have we been able to access so much data at the same time. For some years now, a new generation of algorithms has been able to process this “black gold of the 21st century” faster and faster, making the famous Big Data operate. The discipline that consists of processing this huge mass of data, known as data science, is still in its infancy, but it is already perceived as one of the pillars of the convergence between the ecological and digital transitions.
Sitting in front of his large computer screens, Gilles Faÿ’s eyes are riveted on tables of data and lines of code. He is one of the four members of the team of data scientists at SUEZ’s International Water and Environmental Research Centre (CIRSEE), one of the Group’s skill centres working on innovative resource management. “I started working here two years ago, when the job was created,” explains the expert. “Before that, I was a university lecturer in applied mathematics.” The team could double in size by the end of 2018, demonstrating the keen interest in data science.
In 10 years, the discipline has become increasingly important in resource management strategies, with its predictive scenarios, the identification of faults to optimise yield and the improvement of customer profiles. Data scientists are at the heart of these issues, because they can use a huge mass of information from all sorts of sources, such as meteorological and operational databases, industrial customer and consumer databases, satellite images, CCTV1 videos, etc. SUEZ operates almost three million connected water meters deployed worldwide, which collect consumption indexes several times a day via a radio transmitter.
Data scientists are a rare gem for recruiters, because they possess a range of varied skills, stretching from maths and statistics to programming and even computer engineering.
Making data talk
But Gilles Faÿ does not spend every day working on open source machine learning languages, such as Python or R. He regularly visits “operationals” in the field to understand their needs and expectations. “Once you have formulated the right question, you are halfway there,” he explains.
His job is to answer the question and to find out whether this can be done with data. “Just having the data is not enough,” adds Gilles Faÿ. “It could be just noise, or partial or dirty data.” And this is where the data scientists step in. “Our job is to make the data talk, to clean it up and reorganise and filter it.”
The team at the CIRSEE is not expected to reinvent machine learning, but it uses available algorithms to “put everything to music”. This step still requires professional statisticians, even if the ultimate goal is to train up SUEZ’s business lines, so that they can interpret the results all on their own. “Otherwise, it’s all too easy to make data say any old thing.”
Transforming data for resources’ sake
“SUEZ’s challenge consists of managing data with agility, so that it can all be activated together to reveal the important information,” explains the expert.
After answering the question raised by the business line, the next step is to develop applications using visualisation tools. For example, to create an algorithmic model that detects problems in the water quality distributed in the network (Aquadvanced®) or to optimise the waste collection routes and schedules using volume sensors installed on the bins or weighing information. The classical weekly calendar can be changed according to the volume of waste in the bins.
A number of techniques are used. Clustering, which creates behavioural archetypes, for example, regression, for predictive maintenance or to cut inspection costs, or classification, used to identify leaks on the basis of hydraulic signature classes, etc.
The data to give full customer satisfaction
Benefits can be found in customer relationships, by offering a better price or helping them to reduce their consumption of resources, and in looking for new contracts. Teddo Van Mierle is well placed to explain how. This SUEZ data scientist in the Netherlands has been working on the incorporation of data science into marketing intelligence for nine years. “Today, storage capacity has increased and millions of items of data are available.” Actually, external data (weather and weather forecast, business trends, development information, etc.) is used to enrich SUEZ internal data, achieving a better picture of what is going on and more important, of what are the external and internal influencers on customer behaviour. “So, we have developed new automated platforms that process this data in a few seconds.”
And this is how the Sales Excellence App, developed by SUEZ’s teams, puts all this collected and merged data together and with near real-time high-performance analytics, gives the sales teams, on daily basis, insights on which customers/prospects are most likely to sign a contract, resign and which services are most likely demanded.
Another example is the datalake for SUEZ Trading. “Data from nine European countries is collected and automatically standardised. Then thanks to (predictive) analytics, we can have a complete insight on trends and development of the SUEZ Trading material networks. And thanks to the huge amounts of customer behavioural data gives, SUEZ has, in the end, reliable customer process benchmark information (waste related) for all our customer types. In this way, we can propose service optimisation while increasing our margin. This is where the computer takes over the work, making real-time decisions based on business rules and decision-making logic”.
Teddo Van Mierle works with two other people to collect, prepare, standardise, modify, analyse and visualise the data. Most important is the automation (artificial intelligence) of those processes. “The computers do the rest. But we do check that the predictive models are still valid, or whether the algorithm needs to be adjusted.
From operations to customers
Sophie is the name of the artificial intelligence in Sales Excellence Application that works 24/7, doing the jobs that were previously done by people, who spent their time filling tables and copying and pasting data. “The digital revolution has really changed the world of work,” enthuses Teddo Van Mierle. “Operations, which is the space where Gilles Faÿ works, and sales and marketing prospects, which are my field, come together in data science.”
With around 10 projects per year at the CIRSEE, data science is a promising discipline for the Group. “But we have only just started with Big Data,” admits Gilles Faÿ. “We still have a lot to do with sets of data that remain unused.” Even more so, because the development of machine learning still has a long way to go. In his book, Big Data, penser l’homme et le monde autrement, Gilles Babinet predicts that, “If the power of computers continues to rise at this pace for a few more decades, the capacity of projection of these calculations will simply outstretch our powers of imagination.”
In mid-September 2017, SUEZ turned a new page in its digital transformation with the appointment of Meriem Riadi as Chief Digital Officer.
“My mission is structured around 3 dimensions:
- Defining the Group’s digital roadmap, which is built around 4 axes: the digitisation of customer relationship, the performance of assets and operations, individual and collective performance and finally the setting of new business models (market places in waste, smart city, etc.)
- Accelerating, at the Group level, some operational impact topics: in 2018, the focus would be on the operational applications of data and artificial intelligence,
- Developing open innovation: partnerships with tech start-ups, incubators/accelerators, partnerships with universities and ‘intrapreneurship’ initiatives.”
These new objectives will heavily rely on data collection and processing in order to provide customers with new value-added services and to optimise operations. The challenge for SUEZ is to become a “data driven” company to increase its competitiveness and accelerate the resource revolution.
(1) Closed-circuit television.