A novel neural computing model for fast predicting network traffic

Currently existing web traffic prediction models have the shortages of low accuracy, low stability and slow training speed. Aiming at such problems, this paper proposes a new model to predict the network traffic called MRERPM (MapReduce-based ELM Regression Prediction Model). In this prediction model, Extreme Learning Machine is used to accelerate the training speed and improve the accuracy of prediction. Moreover, a distributed cluster is established based on Apache Hadoop to furtherly improve the processing capacity. Experiment results show that MRERM has a large improvement over training speed compared with other models based on K-ELM or SVR, but not at the cost of accuracy