Developing Hybrid Artificial Intelligence Model for Construction Labour Productivity Prediction and Optimization

• Author / Creator

  Sara Ebrahimi

• Construction labour productivity (CLP) is considered one of the most important parameters affecting the performance of construction projects. Therefore, modeling CLP is a crucial step in construction projects. Accurate prediction of CLP helps in effective planning, cost estimating, and productivity improvement before and during construction project execution. Numerous factors affect CLP and cause complexity in predicting and modeling labour productivity. Thus, CLP modeling and prediction are complex tasks, which can lead to high computational cost and overfitting of data. Since a large number of inputs and high-dimensional data may present different problems, such as reduced accuracy and increased complexity, it is necessary to reduce the dimensionality of CLP data and determine the factors that most influence CLP. This can be accomplished using dimensionality reduction methods, such as feature selection. Existing predictive models of CLP do not focus on dimensionality reduction methods appropriately, which causes reduced accuracy of CLP prediction.
  This thesis presents a novel approach to predict and optimize CLP by applying hybrid feature selection (HFS), machine learning models, and particle swarm optimization (PSO) algorithm. HFS methods select the most predictive factors on CLP to reduce complexity and dimensionality of CLP. Selected factors are used as inputs to four machine learning models, namely adaptive neuro-fuzzy system (ANFIS), ANFIS-genetic algorithm (ANFIS-GA), random forest (RF), and artificial neural network (ANN) for CLP prediction. Results show that the RF model obtains better performance compared to the other three models. Finally, the integration of RF and PSO is developed to identify the maximum value of CLP and the optimum value of selected factors. The new hybrid model presented, named HFS-RF-PSO, is a CLP optimization-and-prediction approach that addresses the limitation of existing CLP prediction studies regarding the lack of capacity to optimize CLP and its most influential factors in regard to a construction company’s preferences, such as targeted CLP. Therefore, the main contributions of this thesis include (1) development of an HFS model to select the most predicting factors on CLP; (2) development and comparison of four different predictive models for CLP and identifying the most accurate model; and (3) development of the HFS-RF-PSO algorithm to identify the maximum value of CLP considering the minimum deviation from the targeted CLP value and also finding the optimum value of the selected.
  The proposed HFS-RF-PSO model will help project managers predict, optimize, and improve the CLP value while taking into account the factors that are most predictive of CLP. The results of this thesis and implementation of the HFS-RF-PSO model will help project managers identify causes of low labour productivity, select and prioritize corrective measures to improve CLP. The model will also enable project managers to improve the reliability of predictions.

• Subjects / Keywords

  • Construction Labour Productivity
  • Artificial Intelligence
  • Optimization
  • Predictive model

• Graduation date

  Fall 2021

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-6gte-2w64

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.