A Novel Deterministic Probabilistic Forecasting Framework for Gold Price with a New Pandemic Index Based on Quantile Regression Deep Learning and Multi-Objective Optimization DOI Creative Commons
Yan Wang, Tong Lin

Mathematics, Journal Year: 2023, Volume and Issue: 12(1), P. 29 - 29

Published: Dec. 22, 2023

The significance of precise gold price forecasting is accentuated by its financial attributes, mirroring global economic conditions, market uncertainties, and investor risk aversion. However, predicting the challenging due to inherent volatility, influenced multiple factors, such as COVID-19, crises, geopolitical issues, fluctuations in other metals energy prices. These complexities often lead non-stationary time series, rendering traditional series modeling methods inadequate. Our paper presents a multi-objective optimization algorithm that refines interval prediction framework with quantile regression deep learning response this issue. This comprehensively responds gold’s dynamics uncertainties screening process various including pandemic-related indices, US dollar index, prices commodities. deep-learning models optimized algorithms deliver robust, interpretable, highly accurate predictions for handling non-linear relationships complex data structures enhance overall predictive performance. results demonstrate QRBiLSTM model, using MOALO algorithm, delivers excellent composite indicator AIS reaches −15.6240 −11.5581 at 90% 95% confidence levels, respectively. underscores model’s high accuracy potential provide valuable insights assessing future trends deterministic probabilistic captures new pandemic index sets benchmark volatile commodities like gold.

Language: Английский

Development of Machine Learning Models for Predicting Bubble-Point Pressure of Crude Oils DOI
Prosper Nekekpemi, Michael W. Totaro, Olatunji Olayiwola

et al.

Published: April 29, 2024

Summary Bubble-point pressure is a crucial parameter in reservoir and production engineering the oil gas industry, but its accurate determination through experimental methods both costly time-consuming. Alternative approaches, such as equations of state empirical correlations like Al Marhoun, Dokla Osman, Glaso, Standing, Vazquez Beggs, are commonly used suffer from limitations including their inability to capture complex, non-linear relationships adapt new or high-dimensional data. This study aims address these shortcomings by developing evaluating range machine learning models—including Decision Tree, Linear Regression, Random Forest, Support Vector Regression (SVR), K-Nearest Neighbors (KNN), AdaBoosting, Gradient Boosting, Stacked Super Learner, Multilayer Perceptron Neural Network (MLPNN)—for predicting bubble-point function temperature, gravity, solution gas-oil ratio, gravity (API). Utilizing comprehensive dataset derived different published papers, total 776 data sets were this which divided into 80% for training 20% testing. The employed performance metrics Average Percentage Relative Error (APRE), Absolute (AAPRE), Root Mean Square (RMSE), Coefficient Determination evaluation. Boosting model emerged most effective, with an RMSE 364.027 R2 0.924 on test data, outperforming existing study. results demonstrate potential models, particularly model, offering advantages capturing complex thereby contributing more effective management strategies.

Language: Английский

Citations

0
Do OPEC+ policies help predict the oil price: A novel news-based predictor DOI Creative Commons
Jingjing Li,

Zhanjiang Hong,

Lean Yu

et al.

Heliyon, Journal Year: 2024, Volume and Issue: 10(14), P. e34437 - e34437

Published: July 1, 2024

The OPEC+, composed of the Organization Petroleum Exporting Countries (OPEC) and non-OPEC oil-producing countries, exerts considerable influence over global crude oil market. However, existing literature lacks a comprehensive application this factor in price forecasting, primarily due to complexity measuring such policy evolutions. To address research gap, study develops news-based OPEC+ index based on text mining methods for analysis forecasting price. First, by crawling news headlines related production decisions, dynamic high-frequency (weekly) is established. Second, linear nonlinear relationship between proposed WTI futures thoroughly examined, assessing potential predictive power explaining movements Third, efficacy constructed rigorously evaluated across eight econometric machine learning models. Key findings include: (1) weekly demonstrates strong concordance with change exhibiting notable peaks troughs corresponding Ministerial Meetings. (2) association (3) For prediction, models incorporating our demonstrate superior performance compared without index. In particular, exhibits more significant effect within three-week horizons performs exceptionally well during periods pandemic Russia-Ukraine conflict. addition, also daily natural gas price, further confirming robust powerful capability energy system.

Language: Английский

Citations

0
Topic-sentiment analysis of citizen environmental complaints in China: Using a Stacking-BERT model DOI
Junling Liu, Ruyin Long, Hong Chen

et al.

Journal of Environmental Management, Journal Year: 2024, Volume and Issue: 371, P. 123112 - 123112

Published: Oct. 31, 2024

Language: Английский

Citations

0
An innovative model to mitigate the impact of oil and steel price dynamics on the oil & gas sector projects DOI

Aguinaldo Júnio Flor,

L. M. Franca

Published: Nov. 17, 2024

This paper addresses the development and application of an innovative model to analyze historical price series commodities, significantly impacting profitability Brazil’s oil gas projects. The experiment focuses on six commodities critical significant exploration companies. It highlights volatility steel prices in Brazilian international markets their direct impact key suppliers explorers sector. research introduces advanced model, employing Deep Learning techniques with automated hyperparameters optimize selection most effective for each dataset. is based a score seven distinct metrics, ensuring choice suitable predict market trends relevant Oil Gas

Language: Английский

Citations

0
A Novel Deterministic Probabilistic Forecasting Framework for Gold Price with a New Pandemic Index Based on Quantile Regression Deep Learning and Multi-Objective Optimization DOI Creative Commons
Yan Wang, Tong Lin

Mathematics, Journal Year: 2023, Volume and Issue: 12(1), P. 29 - 29

Published: Dec. 22, 2023

The significance of precise gold price forecasting is accentuated by its financial attributes, mirroring global economic conditions, market uncertainties, and investor risk aversion. However, predicting the challenging due to inherent volatility, influenced multiple factors, such as COVID-19, crises, geopolitical issues, fluctuations in other metals energy prices. These complexities often lead non-stationary time series, rendering traditional series modeling methods inadequate. Our paper presents a multi-objective optimization algorithm that refines interval prediction framework with quantile regression deep learning response this issue. This comprehensively responds gold’s dynamics uncertainties screening process various including pandemic-related indices, US dollar index, prices commodities. deep-learning models optimized algorithms deliver robust, interpretable, highly accurate predictions for handling non-linear relationships complex data structures enhance overall predictive performance. results demonstrate QRBiLSTM model, using MOALO algorithm, delivers excellent composite indicator AIS reaches −15.6240 −11.5581 at 90% 95% confidence levels, respectively. underscores model’s high accuracy potential provide valuable insights assessing future trends deterministic probabilistic captures new pandemic index sets benchmark volatile commodities like gold.

Language: Английский

Citations

1