Training Data and Source Code

This repository contains the source code used to train and evaluate the machine learning models for defect prediction, as described in the accompanying research paper. The code demonstrates the predictive power of three distinct sets of metrics:- Software and History Metrics: Traditional software and project history metrics.<br>- Human Error (HE) Based Metrics: Novel metrics derived from human error principles.- Combined Metrics: A comprehensive set combining both software and HE metrics.本代码仓库包含用于训练和评估软件缺陷预测机器学习模型的源代码，相关内容已在配套研究论文中进行说明。本代码展示了三类不同指标集的预测性能： - 软件与历史指标集（Software and History Metrics）：传统软件及项目历史指标。 - 基于人为失误（Human Error, HE）的指标集：源自人为失误原理的新型指标。 - 组合指标集（Combined Metrics）：整合软件指标与HE指标的综合指标集合。