A Survey of Data-Driven Construction Materials Price Forecasting DOI Creative Commons
Qi Liu,

Peikai He,

Si Peng

et al.

Buildings, Journal Year: 2024, Volume and Issue: 14(10), P. 3156 - 3156

Published: Oct. 3, 2024

The construction industry is heavily influenced by the volatility of material prices, which can significantly impact project costs and budgeting accuracy. Traditional econometric methods have been challenged their inability to capture frequent fluctuations in prices. This paper reviews application data-driven techniques, particularly machine learning, forecasting models are categorized into causal modeling time-series analysis, characteristics, adaptability, insights derived from large datasets discussed. Causal models, such as multiple linear regression (MLR), artificial neural networks (ANN), least square support vector (LSSVM), generally utilize economic indicators predict commonly used include but not limited consumer price index (CPI), producer (PPI), gross domestic product (GDP). On other hand, rely on historical data identify patterns for future forecasting, main advantage demanding minimal inputs model calibration. Other techniques also explored, Monte Carlo simulation, both uncertainty quantification. recommends hybrid combine various deep learning-advanced analysis potential offer more accurate reliable predictions with appropriate processes, enabling better decision-making cost management projects.

Language: Английский

Network Security Situational Awareness Based on Improved Particle Swarm Algorithm and Bidirectional Long Short-Term Memory Modeling DOI Creative Commons
Peng Zheng, Yun Cheng, Wei Zhu

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(4), P. 2082 - 2082

Published: Feb. 17, 2025

With the continuous development of information technology, network security risks are also rising, and ability to quickly perceive threats has become an important prerequisite means cope with risks. Currently, there various types attacks complex attacking techniques, large differences between them have led difficulty collecting recognizing common characteristics attacks. Considering regular temporal fluctuations in attacks, a method for situational awareness, based on enhanced Particle Swarm Optimization Bidirectional Long Short-Term Memory (BiLSTM) model, is proposed. By gathering organizing critical within network, encapsulated Wrapper feature selection algorithm utilized extraction element features. The refined applied optimize parameters BiLSTM enabling rapid convergence enhancing training efficiency, thus effectively identifying categories experimental results show that MAPE proposed model been diminished 0.36%, while sMAPE 2.654%. Additionally, fitting coefficient attains value 0.92. This indicates high level recognition precision exhibited by detecting risk behaviors. Furthermore, contrast traditional CNN neural more compact, which significantly reduces computational overhead allows efficient awareness.

Language: Английский

Citations

0
MVHS-LSTM: The Comprehensive Traffic Flow Prediction Based on Improved LSTM via Multiple Variables Heuristic Selection DOI Creative Commons
Chang Guo, Jianfeng Zhu, Xiaoming Wang

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(7), P. 2959 - 2959

Published: March 31, 2024

In recent years, the rapid growth of vehicles has imposed a significant burden on urban road resources. To alleviate traffic congestion in intelligent transportation systems (ITS), real-time and accurate flow prediction emerged as an effective approach. However, selecting relevant parameters from information adjusting hyperparameters algorithms to achieve high accuracy is time-consuming process, posing practical challenges dynamically changing conditions. address these challenges, this paper introduces novel architecture called Multiple Variables Heuristic Selection Long Short-Term Memory (MVHS-LSTM). The key innovation lies its ability select informative parameters, eliminating unnecessary factors reduce computational costs while achieving balance between performance computing efficiency. MVHS-LSTM model employs Ordinary Least Squares (OLS) method intelligently optimize cost Additionally, it selects through heuristic iteration process involving epoch, learning rate, window length, ensuring adaptability improved accuracy. Extensive simulations were conducted using real data Shanghai evaluate enhanced MVHS-LSTM. results compared with those ARIMA, SVM, PSO-LSTM models, demonstrating innovative capabilities advantages proposed model.

Language: Английский

Citations

2
Cost Estimation and Prediction for Residential Projects Based on Grey Relational Analysis–Lasso Regression–Backpropagation Neural Network DOI Creative Commons
Lijun Chen,

Dejiang Wang

Information, Journal Year: 2024, Volume and Issue: 15(8), P. 502 - 502

Published: Aug. 21, 2024

In the early stages of residential project investment, accurately estimating engineering costs projects is crucial for cost control and management project. However, current estimation in China primarily carried out by personnel based on their own experience. This process time-consuming labour-intensive, it involves subjective judgement, which can lead to significant errors fail meet rapidly developing market demands. Data collection construction challenging, with small sample sizes, numerous attributes, complexity. paper adopts a hybrid method combining grey relational analysis, Lasso regression, Backpropagation Neural Network (GAR-LASSO-BPNN). has advantages handling high-dimensional samples multiple correlated variables. The analysis (GRA) used quantitatively identify cost-driving factors, 14 highly factors are selected as input Then, regularization through regression (LASSO) filter final variables, subsequently into (BPNN) establish relationship between unit 12 Compared using LASSO BPNN methods individually, GAR-LASSO-BPNN prediction performs better terms error evaluation metrics. research findings provide quantitative decision support estimators investment decision-making.

Language: Английский

Citations

2
Housing Cost Prediction from the Perspective of Grey Fractional-Order Similar Information Priority DOI Creative Commons
Zilin Wei, Lifeng Wu

Fractal and Fractional, Journal Year: 2024, Volume and Issue: 8(12), P. 704 - 704

Published: Nov. 28, 2024

In order to predict the cost of construction projects more accurately for cross-sectional data such as housing costs, a fractional heterogeneous grey model based on principle similar information priority was proposed in this paper. The advantages are proved by stability analysis solution. similarity between predicted samples and existing analyzed, distinguished according index information. factors affecting were sorted similarity, with high ranked first. Since influence tend produce project ranking method can effectively utilize help improve prediction accuracy. addition, compared results other models, it is verified that prioritizing obtain accurate results.

Language: Английский

Citations

1
Foreign Exchange Forecasting Models: LSTM and BiLSTM Comparison DOI Creative Commons
Fernando García, Francisco Guijarro, Javier Oliver

et al.

Published: July 4, 2024

Knowledge of foreign exchange rates and their evolution is fundamental to firms investors, both for hedging rate risk investment trading. The ARIMA model has been one the most widely used methodologies time series forecasting. Nowadays, neural networks have surpassed this methodology in many aspects. For short-term stock price prediction, general recurrent such as long memory (LSTM) network particular perform better than classical econometric models. This study presents a comparative analysis between LSTM BiLSTM There evidence an improvement bidirectional predicting rates. In case, we analyse whether efficiency consistent different currencies well bitcoin futures contract.

Language: Английский

Citations

0
A Survey of Data-Driven Construction Materials Price Forecasting DOI Creative Commons
Qi Liu,

Peikai He,

Si Peng

et al.

Buildings, Journal Year: 2024, Volume and Issue: 14(10), P. 3156 - 3156

Published: Oct. 3, 2024

The construction industry is heavily influenced by the volatility of material prices, which can significantly impact project costs and budgeting accuracy. Traditional econometric methods have been challenged their inability to capture frequent fluctuations in prices. This paper reviews application data-driven techniques, particularly machine learning, forecasting models are categorized into causal modeling time-series analysis, characteristics, adaptability, insights derived from large datasets discussed. Causal models, such as multiple linear regression (MLR), artificial neural networks (ANN), least square support vector (LSSVM), generally utilize economic indicators predict commonly used include but not limited consumer price index (CPI), producer (PPI), gross domestic product (GDP). On other hand, rely on historical data identify patterns for future forecasting, main advantage demanding minimal inputs model calibration. Other techniques also explored, Monte Carlo simulation, both uncertainty quantification. recommends hybrid combine various deep learning-advanced analysis potential offer more accurate reliable predictions with appropriate processes, enabling better decision-making cost management projects.

Language: Английский

Citations

0