Dataset of Flaky behaviors in the web visual

Dataset Overview

 Dataset file

 projects.zip

Dataset file

The data in this file can be used to further analyze flaky behaviour issues in web apps and support improvements in test stability. Our dataset file contains 8 sheets, each categorized as follows:

projects-statistics

This sheet records statistical information for different projects, such as project names, description, the number of stars, flaky commits, flaky style-related commits, etc. It provides an overview of all projects we selected.

 all-data

This sheet contains the complete dataset of flaky behaviour cases, including issue link, commit info, flaky description, causes of flakiness, observed fix strategies and framework, status, priority, etc. It serves as the foundation for our analysis.

Top5-code-distribution-web

This sheet presents the top five code distribution related to web projects, calculate the code line percentage. It helps analyze the prevalence of style-related code.

 causes-statistics

This sheet analyzes common causes of Resource Latency, CSS Attributes and Compatibility, Animations and transitions, Element rendering, and Layout configuration. It serves as a basis for root cause analysis.

 fixed-cases and type statistic

The flaky behaviour cases that have been fixed and categorizes them by type. It provides insights into common fixes and their effectiveness.

priority-statistic

It shows that developers addressing priorities (P1 (high), P2 (medium), and P3 (low)) under different flaky causes.

framework-causes

It shows that different frameworks can have an impact on the nature and frequency of flaky behaviour, underscoring the need for tailored mitigation strategies.

project-commits size impact

It shows that the ratio of visual flaky commits to all flaky-related commits based on the commit numbers per project. Due to the nature of UI development, the impact of visual flakiness tends to be overlooked as the project evolves.

Projects.zip

It contains all the projects we selected from the open-source projects, including source files used to calculate style-related code in the project, calculation results and statistics.