Hands-On Machine Learning for Cybersecurity

Supervised and outlier detection with Knowledge Discovery Databases (KDD)

In this example, we will want to use binary data where 1 will represent a not-normal attack:

from sklearn.model_selection import train_test_splity_binary = y != 'normal.'y_binary.head()

The output can be seen as follows:

Out[43]:0    True 1    True 2    True 3    True 4    True dtype: bool

We divide the data into train and test sets and perform the following actions:

X_train, X_test, y_train, y_test = train_test_split(X, y_binary)y_test.value_counts(normalize=True) # check our null accuracy

The output looks as follows:

True     0.803524 False    0.196476 dtype: float64

On using the isolation forest model, we get this:

model = IsolationForest() model.fit(X_train) # notice that there is no ...