| Abstract | In the present scenario, subcontractors contribute to almost 90% of the overall construction works. Hence, it is essential to assess their performances from the beginning till the completion stage of their work for the success of the construction project. This research focuses on identifying different factors at the work initiation and mid-progress stages of the subcontracted works by verifying them by expert opinion method and assess their mid-progress and final performances by using different AI methods in order to develop a prediction model using WEKA software. The model assesses the schedule performance of the subcontractors as “Delay”, “On time” or “Ahead” and quality performances of the subcontractors as “Poor”, “normal” or “Good”. The survey is done with the contractors of Nepali construction industry working in building construction sector. The result demonstrates that subcontractors in Nepal have better schedule performance when they have high score in factors such as preparedness of extreme weather conditions, understanding of scope, prior working experience, workers’ skill level, quantity of workers, coordination and communication with other subcontractors, protection of completed works, subcontractor commitment, monitoring by the manager/owner of subcontracting party and preparedness of workplace difficulty. Similarly, they have better quality performance when they have high score in factors such as experience in the related field, quality of materials, quality assurance plan, preparedness for work difficulty, worker’s skill, intensity of supervision, defect corrective actions, protection of completed works and monitoring by the manager/owner of the subcontracting party. Among the seven classifiers used in WEKA (Naïve Bayes, Logistic, Multilayer Perceptron, SMO, KStar, J48 and Random Forest), Random Forest had the highest accuracy in 2 models and Multilayer Perceptron and SMO had the highest accuracy in each of the remaining models. The models were developed using k-folds cross validation (k=10). In the end, recommendations were proposed for further implementation and study to improve the schedule and quality performances of the subcontractors in Nepali construction industry as well as to improve the prediction model. |