Today, I presented my talk “Identifying problems in software development with data analysis” at JavaLand. Here you’ll find the slides, the demos, additional information and my answers to the questions asked online.

Talk

You can find the slides to my talk (in German) here:

If you didn’t make it (or don’t speak German), there is an very early version of my talk on YouTube in English.

Demos

Here are the demos that I’ve shown (mostly in German):

See the GitHub-Repository for the Notebooks.

Questions and Answers

Which software did you use for the presentation?

For the talk, I misused PowerPoint, of course 🙂 ! For displaying the Jupyter notebook as slides I used RISE.

How do you analyze code that uses Java Reflection?

I haven’t done this yet, but I think I’ll have a look at this in the near future. Thanks for the hint!

How do you measure the real business usage of an application? A tested component with ~80% test coverage doesn’t represent the real usage of a customer.

I haven’t shown test coverage, but an actual coverage measurement from a “real” application (I clicked through the demo application). See my blog post here for details.

Where can I find the knowledge gap analysis?

You can find the first version of it here. I’m currently rewriting this analysis to boil down the code to ~10 lines of code (without visualization details).

 

Further Reading

Unfortunately, I had to skip the slide with the additional information. But here you find more:

Literatur

  • Christian Bird, Tim Menzies, Thomas Zimmermann: The Art and Science of Analyzing Software Data
  • Tim Menzies, Laurie Williams, Thomas Zimmermann: Perspectives on Data Science for Software Engineering
  • Wes McKinney: Python for Data Analysis
  • Adam Tornhill: Software X-Ray

Software

 

Feedback from the Community

Thanks for the overwhelming feedback after the talk! There were also some tweets that seemed so like my talk :-). Thanks! And thanks JavaLand for having me!

print

My talk at JavaLand 2018

Leave a Reply

Your email address will not be published.