By now, you know the different network anomaly detection techniques. We are now going to build our own network IDS with Python, from scratch. The University of California hosted a competition called The Third International Knowledge Discovery and Data Mining Tools Competition, and they provided a dataset called KDD Cup 1999 Data, or KDD 1990. You can find it at http://kdd.ics.uci.edu/databases/kddcup99/kddcup99.html.
The main aim of the competition was building a system that was able to distinguish between bad (attack) and good (normal) connections. Many modern proposals and machine learning solutions were made using the dataset. But as you can see, the dataset is old; the models were not able to detect modern network attacks, in addition to other issues, like data redundancy. A great study called A Detailed Analysis of the KDD CUP 99 Data Set, done by Mahbod...
