Proposal for bandwidth prediction using Hybrid Division Model
Ognjen Cadovski, Velibor Ilić, Bojan Rikic, Branimir Kovačević · 2021
The Bandwidth value is one of the most significant parameters on internet signal quality. Finding the parameters with the most significant influence on the bandwidth value allows us to predict it using Random Forest Machine Learning algorithm, which proves to be the algorithm most adequate for bandwidth prediction. The problem with using the Random Forest model for bandwidth prediction is that different parameters figure in various types of networks. In order to overcome the mentioned shortcomings in performances, we introduce Hybrid Division Model, which is based on Random Forest Machine Learning algorithm. The role of the Hybrid Division model is to create a separate model for each type of network and operator, which will be analyzed in this paper.