A bad software development process leads to wasted effort and inferior products. In order to improve a software process, it must be first understood. Our unique approach in this paper uses code and test changes to understand conformance to the Test Driven Development (TDD) process. We designed and implemented TDDViz, a tool that supports developers in better understanding how they conform to TDD. TDDViz supports this understanding by providing novel visualizations of developers’ TDD process. To enable TDDViz ’s visualizations, we developed a novel automatic inferencer that identifies the phases that make up the TDD process solely based on code and test changes. We evaluate TDDViz using two complementary methods: a controlled experiment with 35 participants to evaluate the visualization, and a case study with 2601 TDD Sessions to evaluate the inference algorithm. The controlled experiment shows that, in comparison to existing visualizations, participants performed significantly better when using TDDViz to answer questions about code evolution. In addition, the case study shows that the inferencing algorithm in TDDViz infers TDD phases with an accuracy (F-measure) of 87

A bad software development process leads to wasted effort and inferior products. In order to improve a software process, it must be first understood. Our unique approach in this paper uses code and test changes to understand conformance to a process. As a case study, we use these changes to understand conformance to the Test Driven Development (TDD) process. We designed and implemented TDDViz, a tool that supports developers in better understanding how they conform to TDD. TDDViz supports this understanding by providing novel visualizations of developers’ TDD process. We analyze these visualizations using the Cognitive Dimensions framework to discuss findings and design adjustments. To enable TDDViz’s visualizations, we developed a novel automatic inferencer that identifies the phases that make up the TDD process. We evaluate TDDViz using two complementary methods: a controlled experiment with 35 participants to evaluate the visualization, and a case study with 2601 TDD Sessions to evaluate the inference algorithm. The controlled experiment shows that, in comparison to existing visualizations, participants performed significantly better when using TDDViz to answer questions about code. In addition, the case study shows that the inferencing algorithm in TDDViz infers TDD phases with an accuracy of 87%.