3.2.2. Effectiveness Results

We compared the effectiveness using different metrics—accuracy, sensitivity (also named macro recall), specificity, macro precision and macro F1-score [29]. This last metric measures the relation is the harmonic mean of macro precision and macro recall.

$$Specificity = \sum\_{\mathfrak{c}} \frac{TN\mathfrak{c}}{TN\_{\mathfrak{c}} + FP\_{\mathfrak{c}}}, \mathfrak{c} \in clases\tag{1}$$

$$Precision\_m = \sum\_{\mathcal{c}} \frac{TP\_{\mathcal{c}}}{TP\_{\mathcal{c}} + FP\_{\mathcal{c}}}, \mathcal{c} \in \text{classes} \tag{2}$$

$$Recall\_m(sensitivity) = \sum\_{\mathfrak{c}} \frac{TP\_{\mathfrak{c}}}{TP\_{\mathfrak{c}} + FN\_{\mathfrak{c}}}, \mathfrak{c} \in \text{classes} \tag{3}$$

$$F1-score\_m = 2 \ast \frac{precision\_m \ast recall\_m}{precision\_m + recall\_m} \tag{4}$$

where *m* index refers to macro metric and *classes* = {*Pronator*, *Neutral*, *Supinator*}. The term *TPc* refers to the number of samples with class *c* that were classified correctly as *c* by the trained model. *TNc* denotes the number of samples with a different class of *c* that were not classified as *c*. The term *FPc* determines the set of samples with different class of *c* that were classified as *c* by the model and, finally, *FNc* refers to the set of samples with class *c* that were wrongly classified as other different class.

The results obtained with each architecture are shown in Table 2. As can be seen, the reduction in the number of nodes not only maintains effectiveness, but also improves when compared to larger models. The greatest effectiveness is achieved with three nodes in the hidden layer. This may be due to the fact that this hidden layer reduction diminishes the so-called over-fitting phenomenon [30], preventing the model from adjusting too closely to the particular characteristics of the used training subset. The architecture, however, can assimilate enough footprint characteristics even with only two hidden nodes with slightly worse effectiveness.

**Table 2.** Metrics results with different numbers of nodes in the hidden layer. The model trained with only one node in its hidden layer is not able of distinguish one footprint class, so specificity and sensitivity cannot be obtained for this case.

