Onion Analysis

Introduction

In order to analyze community structure we rely on Onion model. All panels related to community structure are based on the onion study included in grimoirelab-elk project.

Thus, structure is provided by dividing contributors into three main groups:

  • Core: those contributing 80% of the activity. These are the most committed developers, and those on which the project relies most.
  • Regular: those contributing the next 15% of the activity. These are people committed to the project, and most likely to become part of the core group or maybe were already in it. The core and regular teams together account for 95% of the activity.
  • Casual: those contributing the last 5% of the activity. There are people in the periphery of the project. However, they are important because it is very likely that future core and regular contributors will come out from this group.

In most models of FOSS development, where there are employees, they usually start directly in regular or core, depending on their positions, experience and responsibilities in the company. On the other hand, non-employees generally start as a part of the casual group. Some of them will become regular and maybe core contributors as they gain experience about the project.

Implementation Details

To better understand how panels are built we need to understand how Onion metric is calculated. We use different levels of granularity to compute onion for different groups of contributors:

  • Globally: takes all data into account, i.e., computes onion for each and every contribution, no matter what organization or project authors belongs to. It is denoted as _Global_ in panels to avoid collisions with existing organization or project names.
  • By organization: splits data by organization and compute onion for each one.
  • By project: same as above, but splits data by project.
  • By organization and project: splits data by organization and project, so onion roles are associated to a person in an organization and a project.

All this data is pre-computed and stored for performance reasons.

Each item in our current onion indices contains fields specified at study schema definition in grimoirelab-elk project.

More details on study configuration can be found at the onion study documentation included in grimoirelab-elk project.

Edit this doc