Predicting Construction Labor Productivity Using Artificial Neural Network

Construction labor productivity is declining
continuously all over the world due to the absence of
systematic data on production rates and; their influential
factors and inadequate prediction techniques. Artificial
Neural Network (ANN) has found to have dynamic learning
and recognition capabilities to effectively predict the labor
productivity in the field of construction management.
Therefore, in this study ANN model has been developed for
predicting the labor productivity focussing on concreting of
floor beams. Selection of relevant influential factors is being
carried out through questionnaire survey. The selected
factors include rainfall, availability of material, maintenance
of equipments, location of project, working space, workforce
skill, level of communication, number of workers, quantity
of concrete, and floor height. Data on production rates and
selected influential factors have been collected from forty
one projects from another questionnaire survey.
Performance of the model has been determined through
calculating the Mean Square Error (MSE). Later, the
performance of ANN model was compared with the result of
Multiple Linear Regressions (MLR). Since it has been found
that ANN predicted the rates more efficiently with least
MSE as compare to MLR. Hence, the influence of rainfall,
availability of material, maintenance of equipments, location
of project, working space, workforce skill, level of
communication, number of workers, quantity of concrete,
and floor height has been effectively incorporated in the
production rates of concreting of floor beams by ANN. For
future projects, production rates for concreting of floor
beams can be effectively predicted by ANN in the presence of
these influential factors.