Open macluo opened 9 years ago
The following SQL will output a list of dates and the commits on that date:
select STRFTIME('%Y-%m-%d', datetime(time,'unixepoch')) as day, count(*) as numCommits from commits
GROUP BY STRFTIME('%Y%m%d', datetime(time,'unixepoch'))
ORDER BY STRFTIME('%Y%m%d', datetime(time,'unixepoch')) ASC;
Going to make some time/commits graphs, not sure if it'll be easier in matlab or not.
I have made the graph for each group and will post them here later.
Oh, cool, that saves me some work hah
The commit history for each group can be viewed in the following image. The bad smells are the ones that make more commits toward the end. I guess second derivative graph will reveal that.
The area under the curve was calculated for each group and compared to the "ideal" curve. The result is the percentage of effort execution. The higher bound is 99% and the lower bound is 54%. Should we say that lower than 66% is bad? It means 1/3 of the effort the group did not utilize.
That sounds good, I'll utilize that.
The timeline of commit history also reveals the work pace of how the actual project has been worked on. The timestamp vs commit plot for most of the groups are somewhat linear. The most linear of all is group 9. The possible bad smells include long hiatus between commits and excessive number of commits toward the end.