Question regarding usage

[gaizka-perez]

Hi team!

I have a quick question regarding the status of the usablity of the Mobgap implementation. I know that you have a disclaimer on the repository regarding the algorithms not being ready for actual research use. We are currently using the python implementation to analyze IMU data and we were thinking on sharing the results in case we found something interesting with the analysis.

Is the warning still relevant for all algorithms (we are thinking on using the healthy pypeline) or are there some of them that would be appropriate to use?

Alternatively, should we look into the Matlab implementation in case we are interested in obtaining meaningful results?

Thanks in advance for your response and for the amazing work you are doing!

[AKuederle]

Hi, The disclaimer is there and still valid, as we haven't completely finished the re-validation and testing of the pipeline. From the tests that we did, algorithms that we tested are on par or better than the old implementation, however, we might still some things before the 1.0 release. For example since the last release (0.10.0) we already made some modifications to the GSD algorithms to improve their performance in some cases. The timeline for the re-validation and the next release (which will likely be a "beta release" followed by a 1.0 soon after) is end of March. I think it is fine to share results that you have. I would suggest you just mention what version you used exactly. If you can wait until the beta release, I would do that and rerun results with the beta version. Then I would be very confident, that there will not be any major changes coming. Hope that helps! We are working to get the beta release out as soon as possible. You can follow the changes in the main branch if you are interested :)

…

On Mon, 10 Feb 2025, 15:43 gaizka-perez, ***@***.***> wrote: Hi team! I have a quick question regarding the status of the usablity of the Mobgap implementation. I know that you have a disclaimer on the repository regarding the algorithms not being ready for actual research use. We are currently using the python implementation to analyze IMU data and we were thinking on sharing the results in case we found something interesting with the analysis. Is the warning still relevant for all algorithms (we are thinking on using the healthy pypeline) or are there some of them that would be appropriate to use? Alternatively, should we look into the Matlab implementation in case we are interested in obtaining meaningful results? Thanks in advance for your response and for the amazing work you are doing! — Reply to this email directly, view it on GitHub <#207>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACSHBO4JJ6DD4ZN5ME2ULD32PC3JNAVCNFSM6AAAAABW2X7UTSVHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZXHE2DCNBVHA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[gaizka-perez]

This is great to know, thanks for your quick response!