session_history.csv - Build history for a project over the course of several months, includes overall build times and status
Problems:
Implement JS app that reads the session_history.csv file, and uses a JS charting library of your choice to present the following time-series:
- passing and failing builds per day, stacked-chart
- build time vs. time, line
Extend the app from (1) to annotate days in the chart that have an "abnormal" number of failing builds. Define (and justify) how you are determining "abnormal".
The input file is CSV formatted with one line per build ("session"). For example:
"session_id","started_by","created_at","summary_status","duration","worker_time","bundle_time","num_workers","branch","commit_id","started_tests_count","passed_tests_count","failed_tests_count","pending_tests_count","skipped_tests_count","error_tests_count"
"951155","[email protected]","2014-09-10 05:38:55 UTC","passed","605.0","10173.0","189","24","production","a82bfc7a1cf085bd72d99651cac5b6c563846581","0","311","0","3","0","0"
"950998","[email protected]","2014-09-10 02:07:17 UTC","passed","674.0","11645.0","197","24","production","b4ac608381bb216ff98366009bbee647eae948aa","0","311","0","3","0","0"
"950659","[email protected]","2014-09-09 23:29:39 UTC","error","0.0","0.0","0","0","production","a662b92fba90e0398a6c47b2db99307c1c60593b","0","0","0","0","0","314"
The interesting columns are:
(3) - Build created_at
(4) - Build final status
(5) - Build overall duration, in seconds
Only copy this repository and open index.html file. Then open session_history.csv via index page.
I have calculated middle value for all falls and found abnormal falls dates.
Middle value calculated by formula:
MV = (summ for all fails per each day)/(count of all fails day)
e.g. I have 10-01-2015 : 10 fails; 11-01-2015 : 3 fails; MV = 13/2 ~ 8;
Then ubnormal here are values than fails count is more than middle value.
10-01-2015 is ubnormal fails day for my example.