Last May, the French retailer Intermarché launched its data challenge : to design a predictive model to estimate the daily sales volumes of several outlets for several products. At the end of June, the group announced its winners who will have the privilege to be the guests of a Paris Machine Learning meet-up on data challenges. They will present their models and give feedback on the methods they used to develop their solutions.

A meet-up to know the keys to win a data challenge

After winning this long-term challenge, Romain Ayres (first in the competition), Pierre Nowak (second) and Jacques Peeters (third) will take part in a meet-up entitled “How to win a data competition” on 6 July, starting at 12.30 pm. They will present their models, the methods and techniques they used to build a successful tool and share their feedback. You can register by following this link.

Romain Ayres, Pierre Nowak and Jacques Peeters are the three winners of the challenge launched by Intermarché and have succeeded in producing models that outperformed the 654 other participants in the competition. They respectively proposed 82, 78 and 33 contributions each, while the record of contribution goes to Changyi Song, sixth of the contest with 102 modifications.

An innovative challenge proposed by Intermarché for data enthusiasts

With 2,665 contributions in total and 657 participants, for an average of 4 attempts for each candidate, the data challenge has involved many data specialists, as shown by the podium of the contest:

• Romain Ayres, the grand prize winner, is a Senior Data Manager at ManoMano, a company specialising in DIY, home and garden. His solution is available here.
• The third winner, Jacques Peeters, is Lead Data Scientist for the same company. His final model is available here, as well as the notebook facilitating its understanding.
• The runner-up of the challenge, Pierre Nowak, is a data scientist and co-founder of Prevision.io, an AI platform that allows data scientists and developers to deliver their AI projects with a better return on investment.

Raphael Sourty, PhD Student for the Renault Group and fourth in the competition with 97 contributions, also proposed his solution which can be viewed here.

All the competitors proposed numerous codes to estimate the daily sales volumes achieved by different Intermarché outlets, and this, for various mass consumption products except for variable weight foodstuffs or fresh products. The first three contestants won cheques worth 3,000, 1,500 and 500 euros each.

Translated from Challenge Intermarché : les vainqueurs présenteront leurs modèles prédictifs le 6 juillet prochain