BrendaQiu
commented
1 week ago I appreciate that there are so many tutorials available, but for my purposes, and for our user's purposes, the need is perhaps a simpler quality control, to see if a participant is following task instructions, or if a data file should be included in the final analysis.
I totally agree that MNE could be hard to navigate because of all its available functionalities (most of which would be redundant in our case). Setting aside the unnecessary dependencies, whether the difficulty of using this package is a deal breaker depends on what we want to do with it. My understanding is that the tech options would be leveraged in one or more of our package's components (say file I/O), but the users don't necessarily have to deal with these tech options themselves. We could embed some functions from MNE in our functions and make them more user friendly. But I do agree that for file I/O, MNE is possibly an overkill that doesn't really do well.
Some comments from Bahar: I overall found this package a little difficult to navigate and like it might have a somewhat steep learning curve. I believe that the ideal user for this tool is a researcher who does not only look at eye-tracking, but combines eye-tracking with other neural data such as EEG/MEG. I know Brenda also uses fMRI so perhaps she can weigh in on this better than I can.
My reasoning mainly comes from me working through the introductory tutorials for this package and finding a lot of information and tutorials, I had to spend a good amount of time sorting through information to find what would be applicable. I had to read through 4-5 different tutorial pages to be able to work through this notebook and fill in our to-do's. The "Working with eye tracker data in MNE-Python" tutorial, for example, goes through an example for which there was eye-tracking along with EEG, so I had to 'filter' the information to understand what would be relevant to me. I appreciate that there are so many tutorials available, but for my purposes, and for our user's purposes, the need is perhaps a simpler quality control, to see if a participant is following task instructions, or if a data file should be included in the final analysis.
I believe a more "straightforward" package, i.e., something that would be easy to use after looking through the README on GitHub, will be more appropriate for our users. This may also be better to accomodate users from a variety of levels of experience/expertise. I am happy to discuss if anyone has other perspectives and if there are other factors I may not be accounting for in my reasoning!