We will now use another machine learning approach to detect malicious URLs. Support vector machines (SVMs) are a popular method for classifying whether a URL is malicious or benign.
An SVM model classifies data across two or more hyperplanes. The output of the model is a hyperplane that can be used to segregate the input dataset, as shown in the following graph:
We then import the required packages. The SVM package available in the sklearn package (as shown in the following code) is very useful for this purpose:
#use SVM
from sklearn.svm import SVC
svmModel = SVC()
svmModel.fit(X_train, y_train)
#lsvcModel = svm.LinearSVC.fit(X_train, y_train)
svmModel.score(X_test, y_test)
Once the model is trained with the SVM classifier, we will again load the model and the feature vector to predict the URL's nature using the model, as shown in the following code...
