Open AccessMACHINE LEARNING IN THE PROBLEM OF FORECASTING THE TIME SERIES OF MUF OF SHORT-WAVE RADIO CHANNELS
Author(s) -
N. A. Konkin,
ANASTASIA DMITRIEVNA PASOVA
Publication year - 2021
Publication title -
nauka v sovremennom mire
Language(s) - English
Resource type - Journals
ISSN - 2524-0935
DOI - 10.52013/2524-0935-62-8-7
Subject(s) - term (time) , computer science , series (stratigraphy) , filtration (mathematics) , artificial intelligence , signal processing , time series , raw data , algorithm , regression , machine learning , telecommunications , mathematics , statistics , paleontology , radar , physics , quantum mechanics , biology , programming language
The article is devoted to the creation of an algorithm for long-term prediction of the values of the MPR. The paper analyzes the influence of various methods of processing raw values of the maximum applicable frequency on the results of machine learning algorithms, such as linear regression and XGBoost. As processing techniques, the Savitsky -Goley filtration method and the isolated forest algorithm were used to determine emissions for the daily course of the MPR.