Empirical Software Engineering using Ultra Large Repositories
description
Transcript of Empirical Software Engineering using Ultra Large Repositories
Empirical Software Engineering using Ultra Large Repositories
Mei NagappanSAIL
Photo: Doug Menuez/Contour by Getty Images/Stanford University Libraries
Agenda• Part 1 – Introduction
– Course Overview and Objectives– Student introductions and expectations– Syllabus– Assignment and Project
• Part 2 – Example of an Ultra Large Repository– World of Code– How to access it?
• Part 3 – Example of on ESE study– What we did?– How we did it?
Typical ESE vs ESE in ULR
What can we learn about SE from these Ultra Large Repositories?
Challenges
Mining Sample Selection Analysis Noise
• Syllabus• Project and Assignment• Break
Example Study
How do ratings
evolve?
128K+
Are Most Apps Great ?
NO
Lots of Apps with very few Ratings
128K+
10K+
Most apps are Average
More Raters => Steady Ratings
More Raters => Steady Ratings
Low Local Rating => Stable
High Local Rating => Unstable
More than 1 star drop => Unrecoverable
Dimensions of StudyDesign