The evaluation metric for this part of competition is based on the following:

The accuracy of your hourly Cooling Load Prediction base model measured by the Normalised Root Mean Square Error (NRMSE):

$$ \epsilon = \frac{1}{{a\_{max}} - {a\_{min}}} \sqrt{\frac{\sum_{i=1}^{n} (p_i - a_i)^2}{n}} $$

Where

- Σ is the NRMSE of forecasts
- n is the total number of forecasts, which is the next 30 days of the data in the testing training dataset
- pᵢ is your prediction value of the cooling load demand in kWh, and aᵢ is the actual cooling load in kWh
- amax and amin is the maximum and minimum actual cooling load in kWh

Submissions are evaluated using NRMSE on the animal clinic dataset).

Leaderboard rankings will be determined based on the calculated scores in ascending order, with lower scores achieving higher ranks.