Skip to content
Open access

Predictive Test Case Prioritization in Agile Workflows: A Machine Learning Approach with SMOTE-ENN and PCA

Oct 2026 · Journal of Informatics and Web Engineering · 0 citations

Abstract

Test Case Prioritization (TCP) is an essential part of agile software development, allowing for the identification and execution of critical test cases early in the process to maximize fault detection efficiency. Traditional TCP techniques often depend on static metrics, like code complexity and historical pass/fail rates, that do not conform with the dynamic nature of Agile workflows. To overcome this, the proposed research presents a discriminative machine learning-based predictive prioritization model that incorporates historical defect data, execution history, as well as requirement criticality to improve auspiciousness of TCP. The study assesses a range of machine learning models like Logistic Regression, Decision Trees, Random Forest, and Gradient Boosting across two datasets, namely CarFinTest and MISPriority. Though the target variable in the MISPriority dataset was nicely structured with respect to the feature set, allowing models to reach a predictive accuracy up to 78.63%, the CarFinTest showed a high degree of imbalance in classes along with poor class separability. Through iterative improvements, the addition of SMOTE-ENN to balance classes, dimensionality reduction via PCA, and feature engineering techniques all greatly increased model performance, with Random Forest and Gradient Boosting performing the best. In addition, the proposed research presents an automatic dashboard for preprocessing dataset, training models, and prioritizing them for implementation into an Agile environment for practical usability. Through adaptive machine learning-based techniques, this research connects traditional and dynamic test prioritization methods, realizing a scalable and data-oriented approach compatible with Agile processes. Future work should focus on enhancing real-time prioritization through incremental learning techniques and expanding dataset generalization across diverse software development projects.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.