GGIR release 2.10-1 on CRAN and short summer break

[Vincent van Hees]

Dear all,

Both Jairo and I are not around to respond to questions in the upcoming two weeks. 

However, I just wanted to leave a note that a new GGIR release (version 2.10-1) was just accepted on CRAN. In the upcoming days the installation binaries for Windows and OSx should become available. Similarly a new GGIRread 0.3.0 release was accepted earlier this week, which is relevant for those of you working with Axivity or GENEActiv data. 

For now it may be good to already report the most critical changes:

1. Major bug fix to .gt3x data (ActiGraph) where autocalibration was performed but not actually applied to the data. This would have affected all analysis done on .gt3x data since we started facilitating .gt3x data in version 2.5-4. We are sorry for not spotting this anytime sooner. If you want to keep on not calibrating the data then could turn the autocalibrations off with argument do.cal = FALSE. ActiGraph csv data is not affected. Thanks to Iris Willems from Belgium for being persistent in asking us to help interpret her funny looking results for some of her participants. It turned out that those individuals wore a poorly calibrated ActiGraph which heavily biased the results. This demonstrates again that it is important that we all remain critical about the GGIR output and not assume that I, Jairo, or anyone else already checked everything.
   
2. We have deprecated the functionality for file formats .wav (Axivity Ltd) .bin (Unilever Discover) as keeping this code up to date with newer R versions took a lot of time. If you are interested in re-inserting this functionality and helping us to make a plan for ongoing maintenance then please contact us.
3. Argument 'nnights' is no longer needed. GGIR should detect the number of nights automatically.
   
4. The code to read cwa data has much improved. In theory this should not have a significant impact on your GGIR output. The improvements primarily relate to the detection of corrupted data blocks and account for them. Another issue that we are now better able to deal with is the situation where a participant curiously plugs the sensor into their computer. In the past the code did not handle these instances well. It kind of ignored those periods, this has now been address too.

A full technical overview of all the updates since release 2.9-0 can be found here. At the end of the month I will write a more readable narrative blog post again to summarise the major updates, which you will then be able to find here: https://www.accelting.com/updates/.

Best,

Vincent