An algorithm for fast and accurate forecasting of time series, such as exchange rates and weather conditions, has been developed in Novosibirsk. The algorithm was created as part of the work of the Siberian State University of Telecommunications and Informatics (SibSUTI), which is under the management of the Ministry of Digital Development. This algorithm is capable of quickly processing large amounts of data online.
According to Anton Rakitsky, Associate Professor of the Department of Applied Mathematics and Cybernetics at SibSUTI, the algorithm implements a method for forecasting time series with maximum efficiency, ensuring high processing speed and forecast delivery. Previously, a similar method was proposed by mathematician Boris Ryabko, but it required significant mathematical calculations and was very laborious. The new algorithm from SibSUTI simplifies this task and allows achieving high forecasting accuracy.
The team of scientists is currently working on creating a library of algorithms and methods for time series analysis. This library will include various optimized methods that will help in forecasting time series. The scientists have already developed methods for reading and preprocessing data, as well as for working with various alphabets and other necessary tools.
The algorithm is planned to be used, in particular, in 2025 to improve the forecast of gas consumption for the company "Газпром межрегионгаз Новосибирск". The goal of the project is to reduce the forecasting error from 10% to 3%, which is of great importance for the effective management of gas reserves in the gas supply system. The team is working on creating a model that will provide accurate forecasts of gas consumption for different categories of users.
Read materials on the topic:
Neural network became a weather forecaster: Yandex launched OmniCast technology for weather analysis
A system for predicting stock price growth has been developed in Moscow
Perm Polytechnic University finds a way to predict hydrate formation in oil wells
