Hybrid Classification Model for Network Traffic Density Estimation

Traffic congestion is an increasing challenge in almost all parts of the globe. In the US alone, over $300B is incurred due to overcrowding. Existing work in the field of “Network Traffic Density Estimation” is hampered by a lack of efficient methods for estimating network traffic density, as conventional methods have difficulty keeping up with the ever-changing topologies of today's networks. Optimizing network performance, boosting security, and allowing efficient resource allocation in complex network infrastructures all depend on finding a solution to this problem. The main cause of congestion at the network level is the increase in demand beyond the potential of the transport network. The machine learning techniques is proposed for the traffic density estimation. The machine learning technique is executed in 3 stages, such as to pre-process the data, extract the features and classify the data. This work suggests a hybrid model which is the combination of logistic regression and K-Nearest Neighbors. The proposed method achieved less error rate as compared to other ML techniques. The proposed method achieved 96.55% accuracy, 0.2% execution time and 3.45% error rate. Python is executed to visualize the suggested model with different metrics like accuracy, execution time, are considered to analyze the results.