Design and methodology for IAM data collection and machine learning model experimentation

This chapter talks about the design approach of the project. It also explains how the IAM training dataset used for the project was collected, and the cleaning and processing techniques. 

The methodology involved in the design and implementation of this project involved three (3) major steps- the installation and configuration of an IAM server (WSO2) which aimed to depict an enterprise environment to provide a real-life dataset for training. The second step after the server setup was to collect real-life data from the server in the form of logs and perform data processing on the collected data. The final step was the selection of a suitable supervised Machine learning algorithm and providing an experiment based on it. In the server, 15 users were created by the super admin and each user was assigned roles with different privileges as shown in Figure 5 below. The experiment was based on users performing normal activities for normal logs and brute force attacks for malicious logs.