On March, 11th and 12th, Romaric Lancien, head of consulting, will represent MFG Labs at Big Data Paris, in the Palais des Congrès de Paris, to pitch a client case with Aymeric Augustin, SVP Data at Canal+ Group.
This eighth edition is the opportunity for a selection of experts to speech about the technical challenges around data projects. Data has become the black gold to any business, and any sector is involved. From improving the competitivity of a business unit, to optimising a product or a service, data has become a key activity to solve business problems, even the most complex ones. Nevertheless, mastering data involves a know-how and a good execution to create value from specific business problems.
“ When the challenges from machine learning meet big data’s”
In this context, Canal+ asked MFG Labs to deep dive into a business problem : how to detect anomalies on their player application myCANAL. Beyond the technical complexity it represented, and beyond the volume of data to be analyzed, the real challenge was to understand the business problem of our client, and to specify it into both a realistic and easy to implement technical solution. The goal? Make the user experience more fluid.
Romaric Lancien, head of consulting at MFG Labs, gives us a few words about the project and the key points to remember from his speech.
Romaric, could you explain to us the notion of Big Data?
Big data is the possibility for a company to collect and tap into a huge amount of data, of different format, to help them improve the way they produce and interact with their clients. This involves mastering the complete data stream : how to ingest vast amount of data in real time? how to store it? how to process it? how to use it to train machine learning algorithms? How to bridge the last mile, which is the ability to have a clear view on this data and make sure that the right decision is made based on this data?
Can you talk to us about Big Data Paris?
Big Data Paris is a trade show which is held every year in Paris, where all the companies facing big data challenges meet with solution providers and makers. The event was launched 8 years ago, when Big Data was starting to be a reality for companies, and now is a real thing. It’s the opportunity to discover all the new data solutions. We can meet with cloud providers, blue chip companies looking for providers, announcers, service providers, integrators, but also startups with specific solutions. It is the opportunity for visitor to discover new use cases, and to meet with partners that could help them implement those.
Why does MFG Labs want to showcase a case study together with Canal+ during the show?
We worked hand in hand with Canal+ team to find the best solution to their anomaly detection problem, and in the end, enhance the service quality. We tackled together a complex problem with multiple dimensions: data architecture constraints, functional challenges, algorithm testing, to find the best implementation that solves the issue. All these dimensions and the dedicated methodology we put in place are valuable, and we wanted to share it together.
What will we learn about this specific use case?
We had the chance to work with the SVP Data of Canal+ Group, who will share his feedback on stage. It’s an opportunity because the topic is common to many companies, though quite complex. We will share the vision and real feedback from a CTO, the challenges he face and the solutions that were contemplated. The topic? The inherent difficulty to anomaly detection: being able to deal with a massive amount of data in near to real time to detect something which is unknown or not easy to labellize upfront. We will talk about the various mathematical approaches we experimented. You need to realise that we are talking here about analyzing 140 millions of logs everyday (1 log = 1 action on 1 device), on streaming (data are coming in real time and must be analyzed directly). The objective is to make the myCANAL app more reliable and fluid for the user.
What value did you bring to your client?
We help them with several things: technical expertise, the ability to mathematically formulate a problem, expertise on the algorithms to use, and on the data engineering part, the ability to manipulate data and AWS services to implement the solution. The consulting part was also huge, because the issue was deeply technical and we could not specify upfront a bullet proof solution. We had first to specify what was the issue about. As a consulting firm and data experts, we have to bring a pragmatic vision to our client’s projects. When we worked out the solution, we tested it thoroughly and set up a validation protocol. Only then were we able to go to the industrialization phase.
One last word?
Be there on March, 12th at 2:40pm in the Blue amphitheatre at Palais des Congrès de Paris!
You can find the presentation deck here.