Figure 8
The figure consists of two side-by-side line graphs and a matrix figure as the third. The left graph is titled “R N N underscore units equals 16 underscore epochs equals 20 – Loss”. The horizontal axis is labeled “Epoch” and ranges from 0 to 6 in increments of 1 unit. The vertical axis is labeled “Loss” and ranges from 0.30 to 0.65 in increments of 0.05 units. Two lines are shown: one labeled “train underscore loss” and one labeled “val underscore loss”. The training loss starts high near 0.67 at epoch 0, decreases sharply to about 0.42 at epoch 1, then drops further to around 0.31 by epoch 2 and remains nearly flat through epoch 6. The validation loss begins around 0.61 at epoch 0, declines to about 0.39 at epoch 1, then slightly increases and stabilizes near 0.40 from epoch 2 through epoch 6. The right graph is titled “R N N underscore units equals 16 underscore epochs equals 20 – Accuracy”. The horizontal axis is labeled “Epoch” and ranges from 0 to 6 in increments of 1 unit. The vertical axis is labeled “Accuracy” and ranges from 0.60 to 0.90 in increments of 0.05 units. Two lines are shown: one labeled “train underscore acc” and one labeled “val underscore acc”. The training accuracy starts near 0.61 at epoch 0, rises sharply to about 0.89 at epoch 1, and then increases slightly to around 0.91 by epoch 2, remaining stable through epoch 6. The validation accuracy starts at 0.72 at epoch 0, increases to about 0.86 at epoch 1, and then remains nearly constant at that level from epoch 2 through epoch 6. The third figure displays a confusion matrix titled “Confusion Matrix – R N N underscore units equals 16 underscore epochs equals 20”. The horizontal axis is labeled “Predicted Label” with categories “0” and “1”, on the left and right. The vertical axis is labeled “True Label” with categories “0” and “1”, from top to bottom. The top-left cell, corresponding to true label 0 and predicted label 0, contains the value 276 and is shaded darkest. The top-right cell, for true label 0 and predicted label 1, shows a value of 0. The bottom-left cell, for true label 1 and predicted label 0, shows a value of 30. The bottom-right cell, for true label 1 and predicted label 1, shows a value of 0. For each matrix, a scale bar is shown on the right, indicating the magnitude of values, ranging from 0 (white) to 250 (dark blue) in increments of 50. The cells are colored according to the scalebar. Note: All numerical values are approximated.The loss and accuracy and confusion matrix of the best model of RNN (units = 16, epochs = 20)