A comparative analysis between CBR based prediction models and MRA models for high-rise construction delay prediction

Abstract

A large number of construction projects being executed across the world are beset with schedule overruns. Although multiple regression analysis (MRA) has been widely considered as a prediction model, techniques like case-based reasoning (CBR) can also be explored for delay prediction. Although CBR has been previously considered for anticipating the construction costs by many researchers, studies pertaining to determining the construction schedule and the subsequent overrun are rare. The research paper develops a prediction model for determining the construction schedule delay using CBR and compares the observations against those obtained using MRA. Primary construction data was collected for real estate projects being executed and nearing completion in Kolkata, India. Using the observations from the primary data, attribute weights were determined using MRA and ANN. Prediction analysis was performed using MRA and using CBR with the attribute weights input from MRA and ANN. CBR-based models show a certain advantage over conventional MRA prediction models as the overall prediction accuracy obtained using CBR analysis showed a higher prediction accuracy as compared to MRA. Extensive usages of techniques like CBR can prove to be beneficial for construction project stakeholders as an excellent tool for systematically utilizing past experiences in future projects predictions.