Discussing the article: "Neural Networks Made Easy (Part 92): Adaptive Forecasting in Frequency and Time Domains"
Neural networks are easy. Part 92 😅
Always your example codes missing some file Neural Networks Made Easy (Part 92) missing file legendre.mqh
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Check out the new article: Neural Networks Made Easy (Part 92): Adaptive Forecasting in Frequency and Time Domains.
The authors of the FreDF method experimentally confirmed the advantage of combined forecasting in the frequency and time domains. However, the use of the weight hyperparameter is not optimal for non-stationary time series. In this article, we will get acquainted with the method of adaptive combination of forecasts in frequency and time domains.
Time and frequency domain are two fundamental representations used to analyze time series data. In the time domain, analysis focuses on changes in amplitude over time, allowing the identification of local dependencies and transients within the signal. Conversely, frequency domain analysis aims to represent time series in terms of their frequency components, providing insight into the global dependencies and spectral characteristics of the data. Combining the advantages of both fields is a promising approach to address the problem of mixing different periodic patterns in real time series. The problem here is how to effectively combine the advantages of the time and frequency domains.
Compared with the achievements in the time domain, there are still many unexplored areas in the frequency domain. In recent articles we have seen some examples of using the frequency domain to better handle global time series dependencies. Direct forecasting in the frequency domain allows using more spectral information to improve the accuracy of time series forecasts. However, there are some problems associated with direct spectrum prediction in the frequency domain. One of these problems is the potential mismatch in frequency characteristics between the spectrum of known data being analyzed and the full spectrum of the time series being studied, which arises as a result of using the Discrete Fourier Transform (DFT). This mismatch makes it difficult to accurately represent information about specific frequencies across the entire spectrum of source data, leading to prediction inaccuracies.
Author: Dmitriy Gizlyk