Prediction model for pile construction productivity rate utilizing the multiple linear regression technique

Abstract

Productivity is a prevalent requirement in the bridge construction industry. The precision of productivity rate measurement is significantly influenced by the ability to recognize and implement the critical factors that impact the productivity rate. However, the significance of productivity in cost reduction and profit generation is fundamental to every construction industry. Bored piles are critical components in the foundation of transportation bridges. The productivity estimation processes for piles are influenced by a multitude of factors, leading to several challenges for estimators in terms of time and cost. Hence, the present study aims to diagnose these issues and evaluate the rate of productivity in pile construction through the utilization of the multiple linear regression (MLR) Technique. The data for this study was gathered via designated questionnaires, on-site interviews, and telephone inquiries to professionals affiliated with various construction firms. A selection of nine factors that have the greatest influence on the productivity of construction have been identified. These factors are considered autonomous variables that have an impact on the rate of pile productivity. The construction productivity rate, which is impacted by the influencing factors, is the dependent variable. An equal number of 84 questionnaire samples were utilized to construct each of the influencing factors incorporated in this model. The work measurement form was designed to collect real-time primary data from the construction site, six data samples for each factor were obtained from different bridge and overpass projects to verify the effectiveness and performance of the model. The study revealed that the multi-linear regression model has a high level of prediction for productivity, with an accuracy rate of 92.93%. Additionally, the correlation coefficient was determined to be 99%. The results indicated a robust correlation among the independent variables in the constructed model, and the predictions generated by the model matched what was observed.