Development of a Machine Learning Algorithm for Tender Bid Evaluation and Contractor Selection with Comparative Analysis Against Traditional Procurement Scoring Methods

Abstract

The selection of contractors in public procurement is a critical process that significantly influences the success of infrastructure projects. Traditional tender evaluation systems commonly rely on deterministic weighted scoring methods that aggregate financial and technical evaluation criteria to determine contractor rankings. Although these approaches provide structured evaluation frameworks, they often suffer from limitations including subjective weighting procedures, inability to capture nonlinear relationships among contractor attributes, and limited predictive capability regarding project success. This study develops a machine learning–based framework for tender bid evaluation and contractor selection and compares its performance with conventional procurement scoring systems. The proposed model utilizes contractor evaluation variables such as bid price, contractor experience, equipment availability, financial ratios, and historical project success rates to train predictive algorithms capable of estimating contractor suitability scores. Ensemble learning techniques are employed to improve predictive accuracy by combining multiple base learners within a unified evaluation framework. The methodology includes data preprocessing, feature engineering, model training, and validation using classification performance metrics including accuracy, precision, F1 score, and ROC-AUC. Empirical results demonstrate that machine learning models outperform traditional scoring approaches in predicting contractor suitability and identifying potential project risks. The findings show that predictive algorithms such as Gradient Boosting and Random Forest provide higher classification accuracy and more reliable contractor rankings than deterministic procurement scoring systems. The proposed framework enhances procurement transparency, reduces subjective bias in contractor evaluation, and supports data-driven decision-making in infrastructure procurement processes. The study contributes to the advancement of procurement analytics by integrating machine learning techniques with tender evaluation systems and provides practical insights for improving contractor selection in modern digital procurement environments.