engleski

Model for Predicting Construction Time by Using General Regression Neural Network

Construction time is an element of every construction contract. Thus, its prediction is of particular interest. This paper presents a construction time prediction model by using General Regression Neural Network. Key data on a total of 70 constructed buildings have been collected through field studies. Chief engineers of construction companies have been interviewed on contractual and actually incurred construction times, contractual and actual construction costs, type of facilities and construction year. General Regression Neural Network (GRNN) from predictive modelling software named DTREG, as new approach in forecasting, was used for building the predictive model to predict the real construction time. Prediction was very accurate with mean absolute percentage error, MAPE, around 2.19 which means that the error of the model is around 2.19%, the coefficient of correlation between the actual and the predicted values is very high, r = 0.99 (rounded) and the coefficient of determination which measures the global fit of the model R2 is 0.97875 (or 97.88%). This paper contributes to and can be useful for the decision process on planning the construction time in construction companies and in the construction industry in general.

construction time; DTREG software; general regression neural network; predicting

nije evidentirano