Software Analytics with Data Science on Software Data
summary
Argues that data science applied to software data uncovers concrete issues: uses Jupyter Notebook, Python and pandas to load CSV coverage reports, compute per-class coverage ratios, group by package and surface candidate packages for removal; uses jQAssistant and Neo4j to scan code into a graph and query for problems (finds static mutable fields, e.g. ownersIndexes written at a specific class/line). Emphasizes problem-driven analyses, reproducible notebooks, and actionable findings rather than a single universal metric.
AI-written summary · about
chapters (13)
- 0:00Software analytics intro
- 1:04Data science on legacy systems
- 3:56Failed quality dashboard attempt
- 5:22Software analytics defined
- 7:10Software data types
- 9:41Data science for analysis
- 11:47Software analytics tools
- 16:46Notebooks for analysis
- 18:54Demo: Code coverage sizing
- 29:12Graph data science
- 33:09Demo: Graph DB queries
- 36:23Summary
- 37:26Q&A
related talks
37mData Science with Python and R
46mJump-Start Your Data Science Learning Java & Jupyter Notebooks
41mComputer Science, Software Engineering, and the Scientific Method
36mCore of Data Science and Open Source Resources
47mA Practical(ish) Introduction to Data Science
01h 01mWhat Is Data Science?
04h 02mHands-on Session on Data Science for Beginners
01h 53mAnalyze your data in Python