Algorithm 1 $Training\left(X_i\right)$

Input: Training datasets $X_i$

Output: NSGA-III-based OCEEMD–WPT Model

1. Calculate the number of reference points;

2. Generate NSGA-III parameters such as population size and values of the objective functions;

3. Apply non-dominated sorting on the population; while iterations $i\le$ maximum number of_iterations do

4. Apply tournament selection with two parents in terms of probability;

5. Apply crossover between two parents;

6. Apply non-dominated sorting on the population;

7. Associatae the populations with reference points;

8. Apply the niche preservation to select individuals associated with each reference point;

9. Store the niche obtained solutions for the next generation;

10. i = i + 1;

End while

Model←Pareto optimal solutions