Analyses of Software Data and Their Interpretations: A Framework of Information Granules

Data collected from software applications such as issue management systems or version control systems are abstract and require their thorough and comprehensive analysis. Automated issue generation is an understudied area in automated software development despite its effectiveness, safety, and satisfaction which increases developer productivity. Analysis of software data from automated issue generation provides information which could be used by relevant tools or monitored as any other feature in the development process. In this paper, we systematically apply a suite of methods, including clustering algorithms, cluster validity indexes, and information granularity, to generate explainable prototypes using software data from generated GitHub Issues. Among other approaches of data analytics, we employ the principle of justifiable granularity and a method of optimal information allocation. These methods are applied to two dimensional synthetic Gaussian data to illustrate the performance of the methods. The study provides the experimental results using the methods applied to real industrial data coming from the 0pdd software. The resultant groups provide some insights into structure for organising puzzles with similar characteristics.