Performance Evaluation In Internet Of Things Dataset Using Machine

Many Internet of Things (IoT) applications face significant hurdles in terms of data
security. Machine Learning (ML)-based intrusion detection systems (IDS) claim to be effective
and accurate at analysing network data and detecting threats. Our suggested technique, nweighted-univariate feature selection, creates a threshold value that serves as a weight, from
which critical features are extracted and then used to machine learning algorithms like support
vector machine (SVM) and decision tree (DT). These models were trained using the UNSWNB-
15 dataset, which was developed in the Australian Center for Cyber Security's Cyber Range Lab
using an IXIA PerfectStrom tool (ACCS). It has a mix of realistic modern normal and
contemporary network traffic assault characteristics. Accuracy, precision, and recall were used to
evaluate the performance of our suggested model. In DT, the proposed model has a greater
accuracy of 96.4 than SVM, which has an accuracy of 89.1.