Complex Desjardins played host to the 2018 Data Cup on Tuesday night. Hundreds of hackers, coders and other techies were on hand for the occasion. The event began two weeks ago with more than 300 participants who had come together to build data-models on consumer spending habits, based on real transactional information. The objective being to develop a model that could potentially help produce more personalized products and services tailored to the needs of any particular individual.
Organizers say this event is the first of its kind in Canada, where a major financial institution makes sanitized, real data available for analysis in this type of competition. In this case the data used was credit card data from Desjardins, which has had all personal info stripped away, leaving only the behaviour and transactions of each of their roughly 17,000 accounts which were analyzed.
“For us, the Data Cup is about understanding the value of data and what you can do with it” said Ilias Benjelloun, Open Innovation Catalyst at Desjardins Lab, and emcee for the evening. “It’s also about educating people. Understanding how much data is captured and used will only increase the digital literacy of citizens and avoid some surprises, like we saw with the recent Facebook data scandal. There is also a lack of data scientists right now and we are hoping to encourage more people to go into that field. We also believe that coming up with tangible solutions will inspire more organizations to explore working with the community. And of course, helping make the Desjardins brand more fun through such events is also a nice by-product of this competition”.
Finalists were judged by a panel including Sofian Hocine from Microsoft, Caroline Pernelle from Ivado, Mathieu Di Mercurio from Dialogue and Mathieu Dérôme from Desjardins, all of whom evaluated the proposed models for accuracy, technical merit, methodology, and presentation.
GhostsInTheMachine, a team comprised of 3 employees of Desjardins Insurance was declared the winner of the Cup, and the $5,000 prize that comes along with it. The team was made up of Cherif Abda – Statistics Advisor, Serge Oligny – Expert Statistical Advisor and Michael Bordeleau-Tassile – Senior Actuarial Advisor, who joked that he came up with the name as a reference to their own tendency towards innovation within Desjardins.
“The competition was to predict the probability of a client defaulting on their credit card. We were given data to train our model and had to come up with multiple variables on our own. We built almost 3,000 variables, indicators that were used in our modelling techniques” said Bordeleau-Tassile, adding “For Desjardins, they want to be better at predicting where they could lose money. They want to prevent that, and our model helps them find which clients to target for that”.
The 4 other finalists were: ManagedToLogin, Dommies, Hello World and Evovest
Earlier in the evening, Benjelloun hosted a panel on responsible data presented by Desjardins Lab and MTLNewTech. Many large organizations have vast amounts of data, and the discussion centered around making that data more open. In other words, accessible to startups or researchers who would take sanitized and structured data from these organizations and use it to train new artificial intelligence technologies. Benjelloun says “Opening the data creates opportunities. The more people you have looking at a problem or at a data set, the better solutions you will have”. He added that Desjardins invites other companies to join them and use its Data Cup platform if or when they decide to open up some of their own data for AI research purposes.
This story was updated July 25 at 5:17PM