New standard csv file to be processed for GGIR

[pinweichen]

Is your feature request related to a problem? Please describe. No, this is not related to a problem. A new wrist-worn Sony wearable device is used for a research project. I want to create a csv file format that can be device-neutral for all future wearable data. I called this the standard data format. The standard data is saved as a CSV file. I have a simple extra script that cleans any device data into this standardized format. The standard data with the first row as a header and the rest of the data following four columns (timestamp, x, y, z).

Describe the solution you'd like

I would like to collaborate to add this standard file type to the GGIR part 1. I noticed there is a setting for MONITOR$INDEPENDENCE and FORMAT$CSV. I have taken the initiative to modify but having issues in troubleshooting. I have two questions regarding this: (1) I have trouble finding where the MONITOR and FORMAT (e.g., MONITOR$GENEActiv or FORMAT$BIN) are set. Would you mind pointing me to where I can modify these two files for part 1 to use? (2) GGIR scripts are nested scripts, and RStudio has issues debugging or pausing when errors occur with nested scripts. Do you have a way for me to look into the errors? I particularly have trouble troubleshooting the g.getmeta script.

Thank you very much.

Benny

[l-k-]

Benny,

GGIR already supports a device-neutral ad-hoc csv file format, you should just use that, no need to add anything.

To make sure the ad-hoc csv format is used, you have to specify the rmc.firstrow.acc argument. In your case, you will need to specify rmc.firstrow.acc = 1 (GGIR will ignore your 1-row column header).

You also need to specify which column has timestamps in it (using rmc.col.time parameter; in your case it will be rmc.col.time = 1), and which columns have acceleration (using rmc.col.acc parameter; in your case it will be rmc.col.acc = 2:4).

You also need to tell GGIR what format your timestamps are in, using rmc.unit.time parameter. I'm assuming you are using Unix timestamps. If so, you will need to specify rmc.unit.time = "UNIXsec"

Another thing that you need to specify is the sampling rate (frequency) of your data, using rmc.sf

You can read more about the ad-hoc csv format in GGIR documentation here, and you can see an example of a GGIR() call here.

[vincentvanhees]

@pinweichen In relation to your second point:

(2) GGIR scripts are nested scripts, and RStudio has issues debugging or pausing when errors occur with nested scripts. Do you have a way for me to look into the errors? I particularly have trouble troubleshooting the g.getmeta script.

GGIR is not a collection of R scripts, but a collection of functions that depend on each other. This is how R packages and software in general works, you need to test them together.

What I often do is:

• Clone the repo from GitHub
• Create new project in RStudio with the GGIR folder as project root.
• Either do devtools::load(".") to load all functions or

  fns = dir("D:/Code/GGIR/R", full.names = TRUE)
  fns = fns[grep(pattern = "sysdata", x = fns, invert = TRUE)]
  for (i in fns) source(i)

  The second option has the advantage that it also works when you are not in the same project.

• Make sure you do not do library(GGIR).
• Use browser() inside the code where relevant to inspect what is happening inside.
• Run GGIR with do.parallel=FALSE to ease debugging, the code will then stop when it sees browser() or errors if the error happens before browser.

[vincentvanhees]

Will close this issue now as it seems to have been addressed. If you cannot get it to work, please open a new issue with a detailed breakdown of what you tried to get it to work including a detailed description of the csv file format or example file.