How do you define the best time to charge electric vehicles at a terminal?

On March 18, Télécom Paris, the Institut Polytechnique de Paris and Total organized a students hackathon. This one week event was dedicated to two problems related to data, digital and responsible energy.

• Problem 1 — strategic analysis challenge: what action plan for Total’s IT department to reduce its carbon footprint?
• Problem 2 — data challenge, for coders: how to define the best time to charge electric vehicles at a terminal?

This last challenge was organized through Total’s data challenge platform. It was so successful, and the students were so committed that we decided to write about it on our blog.

The objective was to predict the best time to recharge an electric vehicle based on the cost of electricity, the energy mix on the network and the customer’s expectations for their vehicle, thanks to a dataset thoughtfully consolidated by Najate, Yann and Florian from Total Digital Factory.

51 students participated, and submitted 928 contributions, until the end of the night since the last submissions arrived at March 24 at 23:59, deadline for participation.

Congratulations to all the participants, and special kudos to the three teams who reached the podium:

• First place: Jérome Dauba, Nicolas Vigne and Antoine Boulat
• Second place: Nicolas Micaux and Erwan Fagnou
• Third place: Rami El Joubbeh and Bashir Abdel Wahed

To achieve their excellent score, Jérome, Nicolas and Antoine first worked on the quality of the data by cleaning up outliers and filling missing values with the nearest neighbour strategy. After that comes the modelling stage: the rate of renewable energy on the grid is calculated per region and a reference weather station was designated to feed the energy mix prediction algorithm. Three approaches were explored: classical statistical modelling with ARIMA and SARIMA and ensemble models with XGBoost. This latter was trained on regional data separately to predict the cost of electricity and the proportion of renewable energies on the grid locally which made them reach the top of the leaderboard.

The (online — covid!) award ceremony was an opportunity to discuss with the students, and also to explain why enterprise data science is even more exciting than a data challenge. Our experts from the Digital Factory explain why:

See you soon for a new challenge!