Dear all,

Thank you for participating in the AI/ML for 5G-Energy Consumption Modelling Challenge. The competition has been an enormous success thanks to your extraordinary creativity and devotion to technical advancement.

During the webinar and discussions at Zindi and email exchange, observant participants pointed out that some solutions might have violated the competition rule prohibiting future values. Predictors may seemly make better predictions by using the future value explicitly or implicitly as model input. However, those predictors that benefit from future values would be impossible to deploy in practice. To ensure the fairness and validity of the competition, we re-evaluated the solutions. The suspicions were confirmed.

In evaluating this matter, we also found that the efforts of the teams are commendable, and the models are pretty creative even without using future values. Additionally, we took into account the participants' bona fide, as they did not use future values intentionally.

Consequently, two options were provided to these participants: 1) to be dropped from the competition, or 2) to submit a new version in which the usage of future values is removed while keeping irrelevant parts unchanged. Both participants acknowledged that there were unintended usage of future values in their solutions, and made updated submission of codes and result files in less than 48 hours.

Their solutions were thus re-scored, resulting in the following score changes:

• Team 1 (Krishna Priya and Rajat Ranjan): 0.043511 -> 0.06327
• Team 2 (Yisak Bule): 0.052725 -> 0.08390

The leaderboard has been updated accordingly.

We are very grateful to those who stood out, insisted on speaking the truth, and helped protect the integrity of the competition. The two teams’ honest behavior and prompt reaction is appreciated as well.

The final ranking and prizes will be announced soon at COP28, Dubai.

Best regards,

The competition host