Any question answered...

You are reading a single comment by @rj and its replies. Click here to read the full conversation.

•

𝔒𝕨𝖑𝘡 in reply to @rj

Split your projects into cohorts based on when they started. Then figure out for each cohort after 1, 2 .. N months, what proportion were completed. Plot those cumulative distributions and compare. If you want to assess significance then you do a chi-squared test.

•

rj in reply to @𝔒𝕨𝖑𝘡

This still means that if a project is finished in the future, that trend in the past would change. Which I think means that it would work if all of your projects were finished. Could something like a mann kendall test be sensible?

•

𝔒𝕨𝖑𝘡 in reply to @rj

This still means that if a project is finished in the future, that trend in the past would change

It's simpler than that. You just compare completion rate at start + N periods between different start period cohorts.

e.g.

About

Avatar for rj @rj started