Machine Learning Model for Predicting Construction Project Success in Tanzania

The construction projects in Tanzania mainly use the post-implementation evaluation method to determine the project completion success. This traditional method led to the discovery of errors and mistakes in project execution only after the project was completed. These errors and mistakes can be avoided only if the construction project implementation were to be monitored in real time by using machine learning models. The purpose of this paper is to present a machine learning model for predicting construction project success in Tanzania using a random forest ensemble learning algorithm. Research methodology employed quantitative and qualitative methods. Data were collected from 26 regions of Tanzania’s mainland. A total of 100 respondents of which are contractors and consultants for construction projects were involved in data collection process through questionnaires and expert interviews to identify factors influencing construction project success. Generative Adversarial Networks (GANs) were used to expand the dataset to 1082. The model was developed, trained, and tested. The model had an accuracy of 97.5% and was deployed in a web-based application. The contribution of this study is to provide a tool that can be used by practitioners and policy makers to predict construction project success in Tanzania.