Enhancing Stock Price Prediction Using Backpropagation Neural Networks with Adaptive Learning Rate and Momentum Coefficient
Abstract
Stock investment is an investment that, in addition to bringing potential profits, can result in losses if it is not possible to analyze and predict stock prices. Machine learning can be used as a tool for stock prediction, one of which is an artificial neural networks. The Backpropagation Algorithm is an artificial neural network algorithm that is commonly used in pattern recogni- tion. The performance of the Backpropagation Algorithm in forecasting was very good. Backpropagation Algorithm prediction has advantages in terms of accuracy; however, it also has weaknesses: it has a slow convergence rate and requires a long training time. To overcome the weaknesses of the Backpropagation Algorithm, improvements must be made to the algo- rithm, by adding an adaptive learning rate and momentum coefficient. The adaptive learning rate and momentum coefficient are methods that aim to increase the effectiveness of the learning rate and coefficient parameters, where these parameters increase the learning speed of the Backpropaga-tion Algorithm. The existence of an adaptive learning rate and momentum is known to accelerate the learning process without significantly reducing the accuracy. Research has been conducted on the implementation of backpropagation artificial neural networks with adaptive learning rate and momentum correlation (BPALM) in predicting stock prices. The comparison results show that the adaptive learning rate and momentum correlation in the Backpropagation Algorithm can accelerate the learning process in the Backpropagation Algorithm. The Backpropagation Algorithm with an adaptive learning rate and momentum correlation requires 30% less learn-ing time than the conventional backpropagation algorithm. In addition to accelerating the learning process, the Backpropagation Algorithm with an adaptive learning rate and momentum correlation also had a high accuracy value of 98%.
Published
Issue
Section
License
License and Copyright Agreement
In submitting the manuscript to the journal, the authors certify that:
- They are authorized by their co-authors to enter into these arrangements.
- The work described has not been formally published before, except in the form of an abstract or as part of a published lecture, review, thesis, or overlay journal. Please also carefully read SITech's Posting Your Article Policy.
- That it is not under consideration for publication elsewhere,
- That its publication has been approved by all the author(s) and by the responsible authorities – tacitly or explicitly – of the institutes where the work has been carried out.
- They secure the right to reproduce any material that has already been published or copyrighted elsewhere.
- They agree to the following license and copyright agreement.
Copyright
Authors who publish with Computing and Information Processing Letters agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License (CC BY-SA 4.0) that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work.
Licensing for Data Publication
Science in Information Technology Letters use a variety of waivers and licenses, that are specifically designed for and appropriate for the treatment of data:
Open Data Commons Attribution License, http://www.opendatacommons.org/licenses/by/1.0/ (default)
Creative Commons CC-Zero Waiver, http://creativecommons.org/publicdomain/zero/1.0/
Open Data Commons Public Domain Dedication and Licence, http://www.opendatacommons.org/licenses/pddl/1-0/
Other data publishing licenses may be allowed as exceptions (subject to approval by the editor on a case-by-case basis) and should be justified with a written statement from the author, which will be published with the article.
Open Data and Software Publishing and Sharing
The journal strives to maximize the replicability of the research published in it. Authors are thus required to share all data, code, or protocols underlying the research reported in their articles. Exceptions are permitted but have to be justified in a written public statement accompanying the article.
Datasets and software should be deposited and permanently archived in appropriate, trusted, general, or domain-specific repositories (please consult http://service.re3data.org and/or software repositories such as GitHub, GitLab, Bioinformatics.org, or equivalent). The associated persistent identifiers (e.g., DOI or others) of the dataset(s) must be included in the data or software resources section of the article. Reference(s) to datasets and software should also be included in the article's reference list with DOIs (where available). Where no domain-specific data repository exists, authors should deposit their datasets in a general repository such as ZENODO, Dryad, Dataverse, or others.
Small data may also be published as data files or packages supplementary to a research article; however, the authors should prefer a deposition in data repositories in all cases.
