You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be nice if all trigger conditions were also event properties in katydid.
One example for this is the first track max NUP. It is a quantity that psyllid triggers on and that can be histogram-ed in event analysis. But it only partly fulfills this principal, because the trigger doesn't care about first or later tracks, it triggers on any bin with high NUP. (When we made this feature an event property, we were taking data in shallow traps and almost all events consisted of just one track.)
Also, even the current simple trigger implementation has more relevant event properties than just the max NUP and there are probably going to be many more in the future.
Here's the current list of event properties that are not implemented in katydid but directly relevant for the trigger:
max NUP in entire event
time in event of the max NUP bin
second highest NUP in entire event
time distance between the two highest NUP bin
I wouldn't say that we absolutely need this feature now and it is too late for the hopefully last round of trigger optimization anyway. But for the future it would make trigger analysis and trigger efficiency prediction a lot easier.
My suggestion would be, that we let the event properties evolve with the trigger algorithm and if we have a new thing we trigger on, it also becomes an event property in katydid.
The text was updated successfully, but these errors were encountered:
From a discussion I had with Elise:
It would be nice if all trigger conditions were also event properties in katydid.
One example for this is the first track max NUP. It is a quantity that psyllid triggers on and that can be histogram-ed in event analysis. But it only partly fulfills this principal, because the trigger doesn't care about first or later tracks, it triggers on any bin with high NUP. (When we made this feature an event property, we were taking data in shallow traps and almost all events consisted of just one track.)
Also, even the current simple trigger implementation has more relevant event properties than just the max NUP and there are probably going to be many more in the future.
Here's the current list of event properties that are not implemented in katydid but directly relevant for the trigger:
I wouldn't say that we absolutely need this feature now and it is too late for the hopefully last round of trigger optimization anyway. But for the future it would make trigger analysis and trigger efficiency prediction a lot easier.
My suggestion would be, that we let the event properties evolve with the trigger algorithm and if we have a new thing we trigger on, it also becomes an event property in katydid.
The text was updated successfully, but these errors were encountered: