Question on using SMPyBandits on Neuroscience Data

[fspinar]

Hi all,

Thanks a lot for this great package !!! I am a neuroscience PhD student, trying to implement multiple bandit algorithms to explain behavioral data collected from animals and humans.

Unfortunately there are not that many open source resources in the neuroscience community in this space (not that I know of, at least). I was wondering if it would be possible to use SMPyBandits to do things like :

1. Fitting different bandit algorithms to behavioral data. This data will be in the form of dictionary or structure, e.g.: data['choices'] = [0,1,1,0,...] %indicating the chosen arm on each trial of the bandit task data['rewards'] = [...] % indiating the obtained reward on each trial of the bandit task

2) Model/parameter recovery for these fits
   
3) Model comparisons on these fits - information criteria like AIC, BIC, and more, or likelihood ratio tests - possibly integrating SMpyBandits with other python libraries tailored for model fits

4) Plotting the results of models

These, together with an integration of some basic behavioral measures ( calculating performance, switching, etc.) could incredibly help neuroscience community. I have been implementing some of these myself, in Matlab... I just came across your python library.

Happy to chat more !

Thanks a lot, Fatih

[Naereen]

Hello @fspinar Fatih, Thanks for your nice words, it's really sweet.

However, I'm deeply sorry, as my reply will be disappointing: as I'm no longer an active researcher (I'm a full-time teacher now), I don't have any free time to allocate to work again on SMPyBandits or other research software.

Feel free to take inspiration from what I implemented in Python for SMPyBandits, but I won't be able to help more, sorry. Most likely I will also not accept any pull request with significant changes, as this project was part of my PhD and is kinda "frozen" as it is.

Regards from France, @Naereen Lilian Besson

[fspinar]

Hello @Naereen

Sorry, I forgot to respond... Thanks a lot for your kind message. I am sorry to hear that you are not working on it anymore... 😞 But I will be looking at your implementations and hopefully making my own implementations for neuroscience purposes.

Thanks again for creating SMPyBandits, and for your response,

Fatih.