Problem in computing foraging efficiency online

[hanhou]

When computing foraging efficiency, we should exclude all autowater trials and ignored trials. @XX-Yin

[alexpiet]

• It makes sense to exclude autowater trials.
• I can see some logic for including Ignored trials. The mouse's foraging efficiency is low because they are not foraging. But I can also see the logic for excluding them: ignored trials don't give us information about the mouse's choices. It might be good to have some quick simulations to justification that logic

[hanhou]

I think it makes more sense to exclude ignored trials because motivation may be orthogonal to their ability to forage, which can be captured by finish_rate and foraging_efficiency_in_finished_trials separately.

[XX-Yin]

I agree to exclude all autowater and ignored trials.

[rachelstephlee]

I can't seem to find where the foraging efficiency is calculated.

I tracked down auto_train_manager, but I sill don't see an online calculation here.

[XX-Yin]

@rachelstephlee It's here https://github.com/AllenNeuralDynamics/dynamic-foraging-task/blob/5120e492cef2c5a4c51b88ad988083bbf8f3527e/src/foraging_gui/MyFunctions.py#L685

[hanhou]

The latest code is here, and here is how I compute it offline.

Although I'm going to move this function to this "basic analysis" shared repo: https://github.com/AllenNeuralDynamics/aind-dynamic-foraging-basic-analysis

[rachelstephlee]

OK. @hanhou if you can let me know once you move the function then I'll fix the bug. I have some other Stan issues that I'll work on in the meantime.

[hanhou]

OK. @hanhou if you can let me know once you move the function then I'll fix the bug. I have some other Stan issues that I'll work on in the meantime.

Yes, I just merged the PR.