Association rule metrics for files that changed together based on the git history

[JohT]

🚀 Feature: Association rule metrics for Git History

Previously, only the commit count existed for files that co-changed (pairwise) together. Even if this is very intuitive it is biased towards files that change very often (README, pom.xml, ....). Now, a couple of other metrics like Confidence, Lift and Jaccard Similarity from the field of association rule learning are available and integrated in existing and newly added reports. Questions like "which files change together surprisingly often?" can now be answered with CSV reports and even visually in a treemap.

• Calculate association rule metrics for co-changing files
• Add histogram plots for pairwise changed git files
• Add reports for frequently changed files
• Optimize treemap with files that are often changed with others

⚙️ Optimization

• Split experimental git history reports from general ones: Now, the more experimental and exploratory part of the Git History Jupyter Notebook is separated into a second notebook and won't be executed in the pipeline anymore.