This study investigates the application of outlier detection and Dimensionality Reduction techniques to enhance the analysis of Internet of Things cybersecurity datasets, focusing on the NF-ToN-IoT dataset, and evaluates the effectiveness of Principal Component Analysis and t-Distributed Stochastic Neighbor Embedding in visualizing high-dimensional data, combined with Grubbs’s test outlier removal technique. Results indicate that t-Distributed Stochastic Neighbor Embedding outperforms Principal Component Analysis in clustering and differentiating attack types, particularly injection and password-based threats. The Euclidean and Cityblock metrics proved to be the most efficient in terms of computational performance and accuracy. These findings highlight the potential of integrating outlier detection with nonlinear dimensionality reduction to improve intrusion detection systems in Internet of Things environments.