EUVIMEDEUVIMED entdecken
Crossref

Proposing a Quantitative CNNs Model for Multivariate Prediction

Iraqi Statisticians Journal · 2025

DOI: 10.62933/g5mk7247 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Iraqi Statisticians Journal, 2025.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Convolution Neural Network (CNNs) are one of the effective approaches in image processing, pattern recognition, and video analysis, but there is a gap in research related to their application to quantitative prediction of multiple variables. In this article, a new algorithm called Quantitative CNNs is proposed, which is used to predict quantitative data by modifying the network structure instead of being prepared to process images. It can be adapted to work with temporal data or multidimensional quantitative data by processing convolution layers after passing the data through the convolution layers, and using dense layers. During the training step, we use quantitative data as inputs and provide the numerical results that need to be predicted as target values, so the network outputs are quantitative numbers instead of classifications by enhancing the training process employing the shark smell optimization (SSO) method. The method was also applied to a real multi-data set representing natural gas prices, namely spot price, futures price, storage level, daily demand, temperature, shipping and transportation costs, geopolitical events, oil prices, and currency exchange rate (Iraqi Dinar vs. US Dollar). We conclude that the proposed algorithm. The proposed algorithm was effective in estimating quantitative time series data compared to the traditional neural network model, and that the shark smell algorithm which used to improve the prediction accuracy gave more accuracy to the method according to the criteria RMSE and Accuracy. The analysis process was carried out using MatLab version 2023b.

Zitieren

APA

(2025). Proposing a Quantitative CNNs Model for Multivariate Prediction. Iraqi Statisticians Journal. https://doi.org/10.62933/g5mk7247

Vancouver

Proposing a Quantitative CNNs Model for Multivariate Prediction. Iraqi Statisticians Journal. 2025. doi:10.62933/g5mk7247

BibTeX
@article{euvimed2025,
  title = {Proposing a Quantitative CNNs Model for Multivariate Prediction},
  journal = {Iraqi Statisticians Journal},
  year = {2025},
  doi = {10.62933/g5mk7247}
}
Kostenlos starten →