Minimum sample frequency

[nkissel]

Describe the bug When reading in data from a custom csv format that contains minute-by-minute data (rather than in the original sampling frequency), a memory error occurs due to the creation of a large matrix. It appears the code assumes a minimum sample frequency. Is this correct?

To Reproduce Read in summarized data that provides minute-by-minute movement, temperature, and light, rather than in the original sampling frequency. Data were originally collected with sampling frequency of 50Hz, but data were exported to minute frequency (ie 1/60Hz). Inputting a very small frequency results in Error: vector memory exhausted (limit reached?) error due to line 117 in g.calibrate().

1. Sensor brand: GENEActiv
2. Data format: "%Y-%m-%d %H:%M:%S"
3. Approximate recording duration: 14 days
4. Are you using a sleep diary to guide the sleep detection: NO

5. Copy of R command used:

   GGIR(
   do.enmo = T,
   acc.metric = "ENMO",
   windowsizes = c(5, 900, 5400),
   mode=c(1,2,3,4,5),
   datadir = "~/GENEActiv3/onef/trunc/223761_20.csv",  # CUSTOM CSV
   
   studyname = "study1",
   rmc.file = "~/GENEActiv3/onef/trunc/223761_20.csv",
   rmc.nrow = Inf,
   rmc.skip = 0,
   rmc.dec = ".",
   rmc.firstrow.acc = 41,
   rmc.col.time = 1,
   rmc.col.acc = 2:4,
   rmc.col.temp = 7,
   rmc.unit.acc = "g",
   rmc.unit.temp = "C",
   rmc.unit.time = "POSIX",
   rmc.format.time = "%Y-%m-%d %H:%M:%S:500",
   desiredtz = "America/New_York",
   rmc.firstrow.header = 1,
   rmc.header.length = 33,
   rmc.headername.sn = "Device Unique Serial Code",
   rmc.headername.recordingid = "Subject Code",
   rmc.sf = 1/60, # LOW SAMPLING FREQUENCY due to data being on minute scale
   minimumFileSizeMB = 0.2,
   
   outputdir=outputdir,
   do.report=c(2,4,5), 
   nonwear_approach = "2013",
   overwrite = T,
   nonWearEdgeCorrection = T, 
   
   strategy = 1,                  
   hrs.del.start = 0,             
   hrs.del.end = 0,               
   maxdur = 28,                   
   includedaycrit = 0,            
   qwindow=c(0,24),               
   bout.metric = 6,               
   excludefirstlast = FALSE,      
   includenightcrit = 0,          
   epochvalues2csv = TRUE,
   do.imp = T,               
   
   sleepwindowType = "SPT", 
   def.noc.sleep = c(),           
   outliers.only = FALSE,         
   HASPT.ignore.invalid = NA,
   HASIB.algo = "ColeKripke1992", 
   ignorenonwear = T,
   do.visual = F,            
   
   Sadeh_axis = 'Y',
   save_ms5rawlevels=TRUE,     
   save_ms5raw_format="csv",   
   save_ms5raw_without_invalid = FALSE,
   part5_agg2_60seconds=TRUE, 
   minimum_MM_length.part5 = 1,
   timewindow = c('WW', 'OO'),       
   visualreport=F
   )

6. Have you tried processing your data based on GGIR's default argument values? Does the issue you report still appear? YES / NO

Expected behavior Not to throw Error: vector memory exhausted (limit reached)

[vincentvanhees]

The code expects you to provide raw data. Minute-by-minute data can never be raw data, and is most likely data pre-processed by other software. GGIR facilitates some specific formats for non-raw data, e.g. from Actiwatch and ActiGraph.

I am closing this issue now, but feel free to re-open if you think I misunderstood something.