Figure 4.
Panel d contains a horizontal bar chart titled Importance feature graph. The vertical axis lists features in descending order of importance: S E P 15, S E P 3, S E P 14, S E P 2, S E P 16, S E P 17, S E P 11, S E P 9, S E P 19, S E P 13, S E P 7, S E P 1, S E P 12, S E P 8, S E P 18, S E P 5, S E P 10, S E P 6, and S E P 4. The horizontal axis is labelled Importance and ranges from 0.00 to 0.10. S E P 15 has the highest importance at slightly above 0.10, followed by S E P 3 near 0.08 and S E P 14 near 0.07. S E P 4 has the lowest importance at about 0.02. Panel e contains a confusion matrix titled Confusion Matrix Depicting Classification performance. The vertical axis is labelled True Class with rows Cause and Effect. The horizontal axis is labelled Predicted Class with columns Cause and Effect. The matrix values are 18 for true Cause predicted as Cause, 2 for true Cause predicted as Effect, 0 for true Effect predicted as Cause, and 26 for true Effect predicted as Effect. Panel f contains a R O C Curve. The horizontal axis is labelled False Positive Rate and ranges from 0.0 to 1.0. The vertical axis is labelled True Positive Rate and ranges from 0.0 to 1.0. A dashed diagonal reference line extends from the lower left corner to the upper right corner. The model R O C curve begins near a true positive rate of about 0.95, rises to 1.0 at a false positive rate of about 0.10, then remains horizontal at 1.0 until the upper right corner. The legend states R O C curve area equals 1.00.(a) Importance feature graph, (b) Confusion Matrix (c) ROC curve
Source(s): Authors’ own