#Time for change prediction series#

10.Volovich, K.I., Denisov, S.A., Shabanov, A.P., Malkovsky, S.I.: Aspects of the assessment of the quality of loading hybrid high-performance computing cluster.

9.Shimall, T.: Traffic Analysis for Network Security: Two Approaches for Going Beyond Network Flow Data, 16 September 2016.

Chen A Law J Aibin M A survey on traffic prediction techniques using artificial intelligence for communication networks Telecom 2021 2 518 535 10.3390/telecom2040029 Google Scholar Hodgea V Krishnanb R Austina J Polakb J Jackson T Short-Term Prediction of Traffic Flow Using a Binary Neural Network Neural Comput.

#Time for change prediction series#

Lysyak AS Ryabko BY Time series prediction based on data compression methods Probl. Mozo A Ordozgoiti B Gómez-Canaval S Forecasting short-term data center network traffic load with convolutional neural networks PLoS ONE 2018 13 2 e0191939 10.1371/journal.pone.0191939 Google Scholar Cross Ref

Blackwell, Maldon, Mass (1997) Google Scholar 4.Pliska, S.R.: Introduction to Mathematical Finance: Discrete Time Models.Stochastic Models and Statistical Inference. 3.Bosq, D., Nguyen, H.T.: A Course in Stochastic Processes.Christoffersen P Diebold F Financial asset returns, direction-of-change forecasting, and volatility dynamics Manage. Assaf Almog A Garlaschelli D Binary versus non-binary information in real time series: empirical results and maximum-entropy matrix models New J. The proposed procedure of the fast prediction is a simple heuristic rule for predicting the sign of the increment of two neighboring values of a random sequence. The paper proposes the use of some results of the theory of random processes for the estimated fast prediction of increments with acceptable accuracy. This may be unacceptable, for example, when using the prediction of traffic characteristics in telecom networks, when predicting the direction of the gradient in optimization problems, etc. First of all, this refers to a non-stationary process, the prediction of which requires constant retraining of the parameters of the predictors used (for example, the weight coefficients of the neural network) over relatively short time intervals, which leads to time costs.

The essential difference of this problem from the prediction of values (ordinates) of time series implementations is the weak correlation of increments, which, from the point of view of the classical theory of time series forecasting, leads to certain difficulties. Among the tasks of prediction, the task of predicting the signs of increments (direction of change) of the time series process is singled out separately.