Construction Project Estimation with LSTM:Materials, Costs and Timelines

Abstract

In the evolving landscape of sustainable construction, determining the most cost-effective materials while minimizing environmental impact remains paramount. This paper introduces a pioneering methodology utilizing Long Short-Term Memory (LSTM) networks that is used to estimate construction materials, particularly focusing on life cycle costs. Inspired by the evident successes of LSTM networks in various applications such as signal processing and water table depth prediction, we designed a system tailored for the construction industry. Our model integrates data from cost estimation databases, facility management databases, and historical construction data to predict the optimal material solutions for building functional parts. Using a standardized approach to data preprocessing, After being trained on years of construction data, the LSTM-based model was validated against traditional estimation methods. Our findings indicate a superior predictive accuracy in the LSTM model, emphasizing its capability to remember past construction patterns and adjust to intricate construction requirements. Moreover, this research uncovers the potential of LSTM in understanding the dynamic interactions in building materials selection, thereby assisting investors in making more informed decisions that cater to unique building circumstances and environmental sustainability.