Comparative Analysis of Classification Algorithms Using Bot_IoT Dataset

This study shows the comparative analysis of classification machine learning algorithms performed on the Bot_IoT dataset. The Bot_IoT dataset is an omnipresent dataset that contains network traffic data collected from Internet of Things (IoT) devices. The dataset provides useful insights into the characteristics of botnet traffic and can be used to create an environment where there are effective detection and mitigation strategies. This research analysis evaluates several well-known classification algorithms on the Bot_IoT dataset. The algorithms considered in this study include, Decision Trees (DT), Random Forest (RF), Support Vector Machine (SVM), and Naive Bayes (NB). The evaluation process consists of several steps. First, the obtained dataset is preprocessed by handling the left out values, normalizing features, and splitting it into testing and training sets. Then, each classification algorithm is trained based on the training set and the hyper parameters are fine-tuned by using the cross -validation techniques. After training, the performance of each algorithm is evaluated on the testing set using different factors such as accuracy, precision, recall and F1 score. Finally, the research results provide insights on the advantages and disadvantages of different classification algorithms for enabling botnet detection using the Bot_IoT dataset. This study compares the performance of each algorithm based on the accuracy, F1- Score, recall and other evaluation metrics, which allows to identify the most effective algorithm. Additionally, this study analyses the computational complexity and training time of each algorithm to assess their practical feasibility in realtime scenarios. The resultant observation can be valuable for researchers working on IoT security, specifically in the area of botnet detection and prevention. By understanding the performance characteristics of different classification algorithms, stakeholders can make informed decisions when selecting an algorithm for performing botnet detection in IoT environments. Moreover, this analysis contributes to the existing body of knowledge on machine learning-based security approaches and promotes further research in this field