Performance analysis of machine learning-based prediction models for residential building construction waste: Fid-Bau Portal

Performance analysis of machine learning-based prediction models for residential building construction waste

Gulghane, Akshay / Sharma, R. L. / Borkar, Prashant

The process of anticipating the amount of waste that will be produced during construction projects can help in reducing the overall waste. In order to make accurate predictions, it is crucial to estimate the amount of waste generated at each stage of the project. This study aimed to develop prediction models to estimate the amount of construction waste at different stages of the construction project for basic civil engineering materials. Data were collected from 134 construction sites in order to accurately calculate the amount of waste generated during the dynamic nature of construction activities. The collected data were analyzed using the decision tree and K-nearest neighbors’ algorithm, and the neural networks performance was studied by providing gross floor area and material estimation. The accuracy of the prediction models was evaluated using the root mean square method and Mean Absolute Percent Error method. The study results revealed that waste is generated at every stage of the construction project and can significantly impact the cost. The study also observed a pattern in waste generation at typical stages of the project. The prediction model showed satisfactory accuracy with an average RSME value of 0.49, indicating the model can be used for predictions. The combined average accuracy of the decision tree and KNN was found to be 88.32 and 88.51%, respectively. These findings can be used as a reference for the management and utilization of construction waste, and can help reduce the amount of waste generated during construction projects. By predicting the amount of waste accurately at each stage of the project, effective measures can be taken to promote sustainable construction practices.