Uma ferramenta para monitoramento da entropia de mudança e sua relação com métricas de software

Abstract

Context: The change entropy is a measure to indicate how much a software suffers changes over a certain period of time. Researches have shown that the increase in entropy can cause disorder in the development process and may lead to an increase in number of software defects. Given this context, there are no tools support for monitoring the relationship between entropy and different software metrics, such as number of authors who have modified a file, number of commits, authorship and code ownership. Objective: Implement and evaluate a tool that enables the monitoring of entropy and software metrics from projects stored on Github to help developers in project management. Tool: The tool is divided into data extraction, calculation of entropy and metrics, calculation of the correlation of Spearman and data visualization. In data extraction, the data is extracted from the Github API and then the entropy and software metrics are calculated. Then Spearman correlation between entropy and metrics are calculated. In the data visualization step, the change entropy values and the software metrics will be displayed using Treemap, Heat Map and time series. The tool was evaluated by analyzing the views generated from the Angular project data, available on Github. Results: In the analysis of the tool it was possible to find more relevant files of the project, measuring the value of change entropy and generating the visualization of Treemap in different periods of time. Calculating the correlation of Spearman and analyzing the Heatmap view in different periods it was possible to observe which metrics present a higher level of correlation with the entropy value of the files. Conclusions: It was concluded that the tool can be used to find the files which deserve more attention from the developers in the project, due to the high entropy value and metrics correlated with entropy.