Developing construction Labour productivity model using artificial neural networks for building projects of concreting activity in Addis Ababa city

Abstract

We cannot imagine the success of labor intensive construction projects by neglecting construction labor productivity. Therefore, improving construction labor productivity (CLP) is a vital task to increase project quality and decrease project cost and time. There are various modeling techniques adopted for predicting production rates of labor that incorporate the influence of various factors but neural networks are found to have strong pattern recognition and higher accuracy to get reliable estimates. Therefore the overall aim of this study was to develop a model using Artificial Neural Networks (ANN) for estimating the CLP of concreting activity for building projects in Addis Ababa city. The CLP data was collected through questionnaire and direct observation of concreting activity. Both primary and secondary data sources were used. From review of past literatures numerous context specific CLP influencing factors were identified. The most critical influencing parameters were selected by calculating the relative importance index (RII) of questionnaire survey data. The productivity data that was used for further analysis and modeling was collected by direct observation of concreting activity. The strength of relationship between CLP and influencing parameters was analyzed using correlation coefficient results generated by Python. CLP model which represents the output of crew within a certain factors was successfully developed using Python. Five objective and six subjective critical influencing parameters were selected using RII. Crew experience, age of workers, and placement technique are the top three influencing parameters which have a strong relation with CLP with correlation coefficient values of 0.5681, 0.5349, and -0.5227 respectively. The model developed within these influencing parameters have higher capability to predict the output of a labor with coefficient of determination (R2 ) value of 92% and mean squared error value of 0.316%. The CLP model was successfully developed within the identified most critical eleven influencing parameters using ANN’s. Therefore the researcher strongly recommends the application of this model for accurate estimation of productivity. For any interested users the final optimal model can be easily accessed in Python, Excel programmed sheet and mathematical formulas.