**6. Results**

Following the relabeling, we set the number of iterations to the value that enables us to obtain the best scores for each type of neural network. Table 2 shows the final values of the hyperparameters.

The static feature selection was achieved with a manual method consisting in removing the empty features from 25,999 to 3359 features, as well as with Pearson's Correlation, from 3359 to 840 features. Table 5 illustrates the results of the static neural network on the test set.

**Table 5.** Static results.


Table 6 presents the results of the dynamic neural network on the test set.

#### **Table 6.** Dynamic results.


Table 7 shows the results of the hybrid neural network on the test set.


