MetricMiner: Supporting Researchers in Mining Software Repositories

Slides from my presentation of the paper published in SCAM 2013: http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6648195&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6648195

Transcript

1. MetricMiner: Supporting Researchers in Mining Software Repositories Francisco Zigmund Sokol, Mauricio Finavaro Aniche, Marco Aurélio Gerosa {sokol,aniche,gerosa}@ime.usp.br
2. Motivation Starting a study in MSR is hard as it involves expensive tasks: ➔ Download source code repositories. ➔ Extract useful information from raw data. ➔ Calculate metrics. ➔ Typically, it’s required to install and configure tools locally.
3. MetricMiner A web application for supporting common MSR studies steps and possibly making use of cloud computing to scale. http://metricminer.org.br/ https://github.com/metricminer-msr/metricminer
4. MetricMiner
9. Replication study To validate our tool, we replicated a study published by Soetens and Demeyer (2010). Does refactoring improve code complexity? Original study: ➔ 1 Java open source project ➔ 800 commits Extracting data from MetricMiner query interface: ➔ 256 Java projects from Apache ➔ 500k commits
10. Next steps ➔ Improve user interface ➔ Improve scalability - maybe moving to Google BigQuery ➔ Gather more data!
11. Thanks! Questions? sokol@ime.usp.br http://metricminer.org.br/ http://codesheriff.com/