Intelligent techniques for predictive analytics in Agile software development.

Area of Science:

• Software Engineering
• Machine Learning
• Data Science

Background:

• Software development complexity requires advanced project management tools.
• Predictive analytics for issue resolution time estimation improves decision-making and resource allocation.
• The Agile Effort Estimation Software (AgES) dataset from GitHub provides rich features for analysis.

Purpose of the Study:

• To analyze machine learning models for predicting software issue resolution times.
• To evaluate model performance using metrics like MAE, MSE, RMSE, and MdAE.
• To identify the most effective methodologies for issue resolution time prediction.

Main Methods:

• Applied traditional and advanced machine learning models (neural networks, random forests, linear regression).
• Utilized the AgES dataset with features like contributor expertise, issue categories, and components.
• Evaluated models using Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Median Absolute Error (MdAE).

Main Results:

• XGBoost algorithm generally performed best across considered error metrics.
• Comparative analysis included the AgES dataset against existing Agile datasets (TAWOS, Choet et al.).
• Model evaluation highlighted practical implications for real-world software project management.

Conclusions:

• Machine learning offers a powerful predictive tool for software project management.
• Accurate issue resolution time forecasts enable better planning and resource management.
• The study details model training, feature importance, and ML's transformative potential in software development.