Estimate-at-completion (EAC) prediction using Archimedes optimization with adaptive fuzzy and neural networks

Construction companies estimate project costs at the beginning of the project; however, many factors impact the final project cost. Estimate at Completion (EAC) is a critical approach for estimating the final cost based on actual project performance. This paper aims to improve EAC predictions by int...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Automation in construction Ročník 166; s. 105653
Hlavní autoři: Abo Mhady, Ahmed, Budayan, Cenk, Gurgun, Asli Pelin
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier B.V 01.10.2024
Témata:
Adaptive neuro-fuzzy inference systems
Archimedes optimization algorithm (AOA)
Civil engineering
Cost estimation
Machine learning
Neural network
Cost estimation
Neural network
Archimedes optimization algorithm (AOA)
Civil engineering
Adaptive neuro-fuzzy inference systems
Machine learning
ISSN:0926-5805
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Construction companies estimate project costs at the beginning of the project; however, many factors impact the final project cost. Estimate at Completion (EAC) is a critical approach for estimating the final cost based on actual project performance. This paper aims to improve EAC predictions by integrating Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Artificial Neural Network (ANN) with Archimedes Optimization Algorithm (AOA). The integration of the input optimization algorithm aims to optimize the input features and explore the factors that significantly affect EAC. Using 306 data points from 13 construction projects in Taiwan between 2000 and 2007, this paper developed hybrid models and found a significant improvement in EAC estimation compared to ANN and ANFIS. •Four models (two single and two hybrid) were developed for EAC predictions.•Single models were developed using Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Artificial Neural Networks (ANN).•Hybrid models were developed by integrating the Archimedes Optimization Algorithm (AOA) into ANN and ANFIS.•AOA selects key parameters, improving accuracy with fewer input parameters.•The AOA-ANN model outperforms the other models with the minimum number of input parameters.
ISSN:0926-5805
DOI:10.1016/j.autcon.2024.105653