Beginning Anomaly Detection Using

IURP

VNOHDUQPHWULFV

LPSRUW

URFBDXFBVFRUH

SUHGV RFVYPSUHGLFW QRYHOWLHV

PDWFKHV QRYHOWLHV>ODEHO@  

DXF URFBDXFBVFRUH SUHGVPDWFKHV

SULQW $8&^`IRUPDW DXF

Figure 2-54.  The code to generate the AUC score

Chapter 2   traditional Methods of anoMaly deteCtion

---

70

Figure 

2-55

 shows the score. That’s pretty good for an AUC score!

Let’s look at the distribution of predictions in Figure 

2-56

.

Figure 2-55.  The generated AUC score from the predictions on the novelty set

Figure 2-56.  Code to display a graph that shows the distributions for the 

predictions

Chapter 2   traditional Methods of anoMaly deteCtion

---

71

As you can see in Figure 

2-57

, the model ended up predicting more anomalies than 

normal data points, but from what the AUC tells us, it managed to classify most of the 

data entries correctly.

Hopefully by now you will have gained a better understanding of what an  

OC-SVM is and how to apply it. Remember, OC-SVM works well for multi-dimensional 

data (in this case, you had 41 columns after dropping the service column) and can 

be used for