CS545 Assignment 8: Time Series Forecasting with Neural Network
نویسنده
چکیده
In today’s content delivery network (CDN), round trip time (RTT) between clients and servers is the most critical performance metric. Most of CDNs assign clients to servers based on RTT. In this assignment, we first employ neural network approach to forecast future RTT between clients and servers based on the past RTTs. Secondly, we experiment with different parameters such as the number of hidden units, the number of inputs and lambda on two different data sets in order to understand how they affect the forecast accuracy. Thirdly, we explain the possible reasons for the observations we get from the second part. At last, we discuss some findings/open issues and present the conclusion. The remaining sections are organized as follows. Section 2 introduces some background of CND and how we measure the RRT between servers and clients in a CDN. Section 3 focuses on the RTT data preparation that is a necessary step for RTT time series forecasting. In Section 4, we discuss how neural network is implemented to forecast RTT time series. Section 5 presents the possible explanations of the observations we get from the second part. In Section 6, we present open issues and conclusion.
منابع مشابه
Comparison of Neural Network Models, Vector Auto Regression (VAR), Bayesian Vector-Autoregressive (BVAR), Generalized Auto Regressive Conditional Heteroskedasticity (GARCH) Process and Time Series in Forecasting Inflation in Iran
This paper has two aims. The first is forecasting inflation in Iran using Macroeconomic variables data in Iran (Inflation rate, liquidity, GDP, prices of imported goods and exchange rates) , and the second is comparing the performance of forecasting vector auto regression (VAR), Bayesian Vector-Autoregressive (BVAR), GARCH, time series and neural network models by which Iran's inflation is for...
متن کاملOptimal Modeling and Forecasting of Equipment Failure Rate for the Electricity Distribution Network
In order to gain a deep understanding of planned maintenance, check the weaknesses of distribution network and detect unusual events, the network outage should be traced and monitored. On the other hand, the most important task of electric power distribution companies is to supply reliable and stable electricity with the minimum outage and standard voltage. This research intends to use time ser...
متن کاملAN EXTENDED FUZZY ARTIFICIAL NEURAL NETWORKS MODEL FOR TIME SERIES FORECASTING
Improving time series forecastingaccuracy is an important yet often difficult task.Both theoretical and empirical findings haveindicated that integration of several models is an effectiveway to improve predictive performance, especiallywhen the models in combination are quite different. In this paper,a model of the hybrid artificial neural networks andfuzzy model is proposed for time series for...
متن کاملComparative Study of Static and Dynamic Artificial Neural Network Models in Forecasting of Tehran Stock Exchange
During the recent decades, neural network models have been focused upon by researchers due to their more real performance and on this basis, different types of these models have been used in forecasting. Now, there is a question that which kind of these models has more explanatory power in forecasting the future processes of the stock. In line with this, the present paper made a comparison betw...
متن کاملVehicle's velocity time series prediction using neural network
This paper presents the prediction of vehicle's velocity time series using neural networks. For this purpose, driving data is firstly collected in real world traffic conditions in the city of Tehran using advance vehicle location devices installed on private cars. A multi-layer perceptron network is then designed for driving time series forecasting. In addition, the results of this study are co...
متن کامل