Hands-On Artificial Intelligence for Cybersecurity

A phishing detector with logistic regression

We can then use logistic regression to implement a phishing detector, exploiting the fact that logistic regression is particularly useful for solving classification problems. Like spam detection, phishing detection is nothing more than a sample classification task.

In our example, we will use the dataset available on the UCI machine learning repository website (https://archive.ics.uci.edu/ml/datasets/Phishing+Websites).

The dataset has been converted into CSV format starting from the original .arff format, using the data wrangling technique known as one-hot encoding (https://en.wikipedia.org/wiki/One-hot), and consists of records containing 30 features that characterize phishing websites.

Find ...