Evaluation of Factors Delaying the Completion of Schools Projects by Using Neural Network Multi-Layer Perceptron Model

Abstract

Construction project delays represent a persistent challenge in Iraq’s educational infrastructure sector, resulting in financial losses and compromised learning environments. This study presents a predictive framework that combines statistical analysis with a Multi-Layer Perceptron (MLP) neural network to evaluate and classify the causes of delays in school construction projects. Initially, data were collected from (104) valid respondents through structured questionnaires addressing 38 potential delay factors. Relative Importance Index (RII%) analysis identified critical causes such as lack of financial allocations, delays in contractor payments, and ineffective scheduling. Subsequently, an (MLP) model was trained using (70%) of the dataset and tested on the remaining (30%) to classify delay responsibilities among four categories: Owner, Contractor, Consultant, and External Causes. The (MLP) architecture included three dense layers interspersed with dropout layers to prevent overfitting, and performance evaluation was conducted using accuracy, F1-score, Receiver Operating Characteristic (ROC), Area Under Curve (AUC), and confusion matrix analysis. The model achieved a test accuracy of (95.24%), with class-specific F1-scores of 1.00 (Owner), 0.94 (Contractor), and 0.92 (External Causes) with Area Under Curve (AUC) scores reached up to 1.00, confirming excellent discriminative power. The top 15 selected features highlighted financial, technical, and procedural delays, aligning with the statistical (RII%) rankings. These results demonstrate that the (MLP) model offers a robust, data-driven tool for early classification and intervention, helping stakeholders improve project planning, resource allocation, and accountability in school construction delivery in Iraq.