Risk prediction and intervention strategies for civil and commercial contract disputes based on genetic algorithms and machine learning

Contract disputes in civil and commercial projects pose significant financial, legal, and operational challenges, creating a need for reliable data-driven dispute risk assessment methods. The research presents a proof-of-concept predictive and prescriptive analytics framework based on a Genetic Algorithm optimized Weighted Least Squares Support Vector Machine (GA-WLSSVM) for contract dispute risk prediction. The proposed approach integrates adaptive sample weighting with GA-based parameter optimization to improve classification robustness, particularly under class imbalance conditions. Experiments are conducted on a publicly available Kaggle civil and commercial contract dispute risk dataset, consisting of numerical and categorical attributes. The dataset is partitioned using a stratified 80/20 train–test split with fixed randomness to ensure reproducibility. GA optimization is performed on the training subset to determine optimal penalty and kernel parameters, while final performance is evaluated on a held-out test set. The proposed GA-WLSSVM achieves best optimized validation accuracy of 0.94 and F1-score of 0.91. In addition to predictive modeling, exploratory, model-derived analysis of dominant risk factors is used to generate indicative intervention insights, rather than deployment-validated optimization strategies. The results demonstrate methodological effectiveness within the experimental dataset, supporting the feasibility of the proposed approach. However, the empirical evaluation is limited to a single benchmark dataset with restricted attribute diversity, and the intervention analysis is based on model interpretation rather than real-world implementation. Consequently, the findings should be viewed as dataset-specific proof-of-concept evidence, with further validation on independent, real-world, and multi-jurisdictional contract datasets required before practical adoption in operational decision-support systems.