Novel approach to determine release day by addressing stochastic nature of software fault data
Md. Rashedul Islam, Momotaz Begum, Md. Nasim Akhtar, Jia Uddin · Data Science and Management · 2025
Creating software that is free from bugs and errors is challenging. A bug refers to any imperfection or flaw in software that causes it to function incorrectly or unexpectedly. On the other hand, a fault (or failure) can be either a hardware defect or a software/programming mistake (bug). This study analyzed the stochastic behavior of fault occurrences to detect the most optimal day for software release from a fault dataset. In the absence of prior analysis of the randomness of fault occurrences, we investigated the suitability and performance of the metaheuristic approach for identifying the optimal release day. We employed several metaheuristic algorithms to determine the optimal release day and measured their performances by investigating various criteria. The simulated annealing and genetic algorithm outperformed all the algorithms used in this study in determining the optimal software release day by analyzing past fault data. The ideal release day was determined by analyzing fault occurrences in the dataset over various time frames to identify periods when the number of daily faults falls below a specified threshold. Various performance metrics were used to understand the relativity of each algorithm and evaluate its performance in terms of providing the expected output with minimal faults.