The subject of this paper is computer worm detection in a network. Computers worms have been defined as a process that can cause a possibly evolved copy of it to execute on a remote computer. They do not require human intervention to propagate; neither do they need to attach themselves to existing files. Computer worms spread very rapidly and modern worm authors obfuscate their code to make it difficult to detect them. This paper proposes to use machine learning to detect them. The paper deviates from existing approaches in that it uses the darkspace network traffic attributed to an actual worm attack to validate the algorithms. In addition, it attempts to understand the threat model, the feature set and the detection algorithms to explain the best combination of features and why the best algorithms succeeds where others have failed.