Skip to content

Latest commit

 

History

History
34 lines (29 loc) · 3.81 KB

Conversion_Rate_Glossary.md

File metadata and controls

34 lines (29 loc) · 3.81 KB

Conversion Rate and Related Terms Glossary

A set of terms and definitions that I think are useful for the Conversion Rate.

Contribution Tiers: Equivalent to funnel "layers"

Types of Conversion Rates

  • Hierarchical Conversion Rates: Multiple Conversion Rates in a related set, forming a pyramid shape, where each Conversion Rate represents % of conversion to a higher level. These can also be seen as multiple related standalone conversion rates.
    • Example: Group A, D0, D1, D2.
  • Standalone Conversion Rates: One number, consisting of a denominator representing "new" contributors and the numerator representing "sustained" contributors. Different Conversion Rates can be obtained due to the data being examined combined with filters.
    • Examples:
      • Conversion Rates at the repo, project or community levels
      • Conversion Rates with respect to different intervals / different time periods
      • Conversion Rates with respect to different developer roles (D0, D1, D2 for example)
      • Conversion Rates with respect to different types of contributions (technical, non-technical). As mentioned on https://chaoss.community/metric-types-of-contributions/, it would be better to recognize different types of contributions independently.

Definitions of Activity-Based Roles in Conversion Rate: New and sustained are the most relevant here, as they are directly mentioned in the metric's question. Thanks to Elizabeth Barron for suggesting standardizing these definitions - they are kept consistent as possible with other CHAOSS metrics.

Analytical Unit: - Conversion Rate can be calculated at the repository, project, or community level. It can be computed for a single repo/project/community but ultimately, we seek to compute Conversion Rate for pairwise comparison between projects/communities. Therefore, a set of one or more repositories will be the analytical unit (invariant) considered. By adjusting the set of repositories used for data collection, different conversion rates may be obtained. - Choosing Repo for Data Collection - In my approach: for the initial testing of the Conversion Rate Metric Model, I will start with a single active repository to use as data (e.g. Augur), then once the code is stable, analyze a larger collection of repositories as the analytical unit and compare across multiple projects - for example Tensorflow vs. Pytorch.