Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Construction Forecasting Using Recurrent Neural Networks
Despite all their advantages, univariate and multivariate time series models are linear statistical methods subject to significant limitations for characterizing nonlinear relationships. Machine learning models, such as neural networks, have established themselves as a serious alternative to classical statistical models for exploring nonlinear relationships. This chapter introduces recurrent neural networks (e.g., long short-term memory and gated recurrent units) for nonlinear time series forecasting and explains the life cycle of construction time series forecasting using such networks. These networks are designed and trained to forecast Highway Construction Spending (HCS) time series. Also, their forecasting performances are investigated and compared with those of seasonal ARIMA and VEC models. The comparison results show that recurrent neural networks (i.e., long short-term memory and gated recurrent unit networks) can provide higher accuracies in forecasting the long-term variations of HCS than statistical linear time series models based on typical error measures.
Construction Forecasting Using Recurrent Neural Networks
Despite all their advantages, univariate and multivariate time series models are linear statistical methods subject to significant limitations for characterizing nonlinear relationships. Machine learning models, such as neural networks, have established themselves as a serious alternative to classical statistical models for exploring nonlinear relationships. This chapter introduces recurrent neural networks (e.g., long short-term memory and gated recurrent units) for nonlinear time series forecasting and explains the life cycle of construction time series forecasting using such networks. These networks are designed and trained to forecast Highway Construction Spending (HCS) time series. Also, their forecasting performances are investigated and compared with those of seasonal ARIMA and VEC models. The comparison results show that recurrent neural networks (i.e., long short-term memory and gated recurrent unit networks) can provide higher accuracies in forecasting the long-term variations of HCS than statistical linear time series models based on typical error measures.
Construction Forecasting Using Recurrent Neural Networks
Shahandashti, Mohsen (Autor:in) / Abediniangerabi, Bahram (Autor:in) / Zahed, Ehsan (Autor:in) / Kim, Sooin (Autor:in)
Construction Analytics ; Kapitel: 5 ; 75-94
25.04.2023
20 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Geomorphology-based Time-Lagged Recurrent Neural Networks for runoff forecasting
| Springer Verlag | 2012
Geomorphology-based Time-Lagged Recurrent Neural Networks for runoff forecasting
| Online Contents | 2012
| DOAJ | 2019
Early Forecasting of Rice Blast Disease Using Long Short-Term Memory Recurrent Neural Networks
| DOAJ | 2017