Over the years, the use of internet has grown exponentially. As a result, crime on the internet has also grown. Botnets serve as the main technological backbone for a wide array of cyberattacks. As evident from various literatures, machine learning algorithms has a lot of potential in the detection of botnets. However, dimensionality of real-world datasets creates bottleneck in analysis. In this context, feature selection techniques have come up as a great tool in reducing the dimensionality without losing the physical interpretation of the original data. In this paper, we compare three different approaches of feature selection. We explore and compare three feature selection techniques categorised under filter, wrapper, and embedded methods. After conducting feature selection, we have employed six supervised machine learning classifiers for classification and detection of P2P botnet flows. Additionally, we have employed majority voting ensemble learning algorithm to improve the classification results.

Summary Machine learning approaches are widely used for the detection and classification of emerging botnet variations due to their ability to yield more precise results compared to traditional methods. The relevancy of the features plays a major role in these detection algorithms' effectiveness. As such, the most distinctive characteristics must be extracted from a high‐dimensional dataset that is used to classify botnets. Nevertheless, we discovered that the majority of earlier studies lacked proper analysis and paid little attention to the various feature selection techniques. The main goal of this work is to investigate and assess the advantages and disadvantages of the different feature selection techniques used for botnet detection. Studies show that feature selection is a very efficient way to decrease the amount of storage and processing power required while simultaneously increasing classification accuracy. As a consequence, its application in many other fields has grown. The field of feature selection is recognized for its non‐deterministic polynomial‐time hardness; to mitigate this hardness, metaheuristic techniques have been applied. Metaheuristic algorithms are exceptionally good at performing a global search. In order to choose feature subsets optimally in the field of botnet detection, we additionally prioritize the use of metaheuristic methods. This study offers a more thorough insight of the feature selection strategies that are primarily employed by machine learning‐based botnet detection models. It also offers insights into how better feature selection approaches might be applied to strengthen botnet detection mechanisms. Additionally, it will help in understanding the limitations of existing approaches and identifying areas for improvement.

Peer-to-peer (P2P) botnet is one of the greatest threats to digital data. It has become a common tool for performing a lot of malicious activities such as DDoS attacks, phishing attacks, spreading spam, identity theft, ransomware, extortion attack, and many other fraudulent activities. P2P botnets are very resilient and stealthy and keep mutating to evade security mechanisms. Therefore, it has become necessary to identify and detect botnet flow from the normal flow. This paper uses supervised machine learning algorithms to detect P2P botnet flow. This paper also uses an ensemble learning technique to combine the performances of various supervised machine learning models to make predictions. To validate the results, four performance metrics have been used. These are accuracy, precision, recall, and F1-score. Experimental results show that the proposed approach delivers 99.99% accuracy, 99.81% precision, 99.11% recall, and 99.32% F1 score, which outperform the previous botnet detection approaches.