Conversation
|
Says 1 check failed, but that looks bogus (0.00% decrease in coverage) Also a quick note, both the new and old event detection functions scan the 12 hours worth of second data and report significant events in less than an second each. The increase in run time for our new function seems to be negligibly significant. |
|
Those checks currently only report on the landingzone. It would be awesome if someone added some tests for the API! |
|
I tried re-running the Travis build (which reports coverage to Coveralls), but it didn't help. I don't see any errors on either the Travis or Coveralls sites, so I think you can safely ignore this error. |
|
I hope to have this updated and ready to merge by week's end. Don't merge until another commit has been made. |
|
Any update on this? |
|
Ping Any update on this? |
detect_events.py has been changed. It can now stretch to any window size.
Also added ipython notebook (with 12h of data), visible in Github GUI. It goes through an explanation of the detect_events.py, showing how our data is transformed before it is put into the function, as well as how the update to detect_events.py works compared to the old code.
Definitely open to criticism, the new algorithm runs a little bit slower, but is much more flexible. Would like to drop run time even more.
(note: looking to make a cronjob that runs detect() on new data as it comes in for next feature)