Network Traffic Forecasting Based on Logistic Iterative Regression Model

Abstract

The size of the network traffic is of great significance to the design of the network architecture. This paper forecast network traffic based on logistic regression model, proposes an improved network traffic forecasting method. In this method, the logistic regression model parameters need to be estimated from historical data. For the three unknown parameters in the logistic regression model, first use the Neyman-Fisher factorization theorem to obtain the unbiased sufficient statistics of one of the parameters. Under the assumption that the general solution is known, use the least square method to solve the other two parameters. Then, under the premise of satisfying the constraints, the scope of the general solution is determined. Among all the parameters, the parameter with the smallest model error is selected to obtain the logistic regression prediction model. Experimental simulations prove that the method improves the accuracy of network traffic forecasting.