The article discusses the methodology behind an obfuscated malware dataset designed for evaluating detection methods. It consists of 58,596 records equally split between benign and malicious samples, covering three malware categories: Spyware, Ransomware, and Trojan Horse. Data preprocessing steps were applied, enhancing the dataset's quality before conducting two classifications—one for general malware detection and another for identifying specific malware types. The research highlights the dataset's 55 features and the importance of model evaluation in improving malware detection techniques.
The obfuscated malware dataset contains 58,596 records used to evaluate malware detection methods, comprising 50% benign and 50% malicious samples from various malware types.
After preprocessing the data, we conducted two sets of classifications: one for malware detection and one for classifying the types of malware in the dataset.
Collection
[
|
...
]