wearingMarking function

Classify Wear and Nonwear Time for Accelerometer Data

This function classifies wear and nonwear time status for accelerometer data by epoch-by-epoch basis.

wearingMarking(
  dataset,
  frame = 90,
  perMinuteCts = 60,
  TS = getOption("pa.timeStamp"),
  cts = getOption("pa.cts"),
  streamFrame = NULL,
  allowanceFrame = 2,
  newcolname = "wearing",
  getMinuteMarking = FALSE,
  dayStart = "00:00:00",
  tz = "UTC",
  ...
)

Arguments

• dataset: The source dataset, in dataframe format, which needs to be marked.

• frame: The size of time interval to be considered; Window 1 described in Choi et al. (2011). The default is 90.

• perMinuteCts: The number of data rows per minute. The default is 1-sec epoch (perMinuteCts = 60). For examples: for data with 10-sec epoch, set perMinuteCts = 6; for data with 1-min epoch, set perMinuteCts = 1.

• TS: The column name for timestamp. The default is TimeStamp .

• cts: The column name for counts. The default is axis1 .

• streamFrame: The size of time interval that the program will look back or forward if activity is detected; Window 2 described in Choi et al.

  (2011). The default is the half of the frame.

• allowanceFrame: The size of time interval that zero counts are allowed; the artifactual movement interval described in Choi et al. (2011). The default is 2.

• newcolname: The column name for classified wear and nonwear status. The default is wearing . After the data is processed, a new field will be added to the original dataframe. This new field is an indicator for the wearing (w ) or nowwearing (nw ).

• getMinuteMarking: Return minute data with wear and nonwear classification. If the source is not a minute dataset, the function will collapse it into minute data. The default is FALSE.

• dayStart: Define the starting time of day. The default is the midnight, "00:00:00". It must be in the format of "hh:mm:ss".

• tz: Local time zone, defaults to UTC.

• ...: Parameter settings that will be used in dataCollapser function.

Returns

A data frame with the column for wear and nonwear classification indicator by epoch-by-epoch basis.

Details

A detailed description of the algorithm implemented in this function is described in Choi et al. (2011).

Note

Warning: It will be very slow if accelerometer data with 1-sec epoch for many days are directly classified. We recommend to collapse a dataset with 1-sec epoch to 1-min epoch data using dataCollapser and then classify wear and nonwear status using a dataset with a larger epoch.

Examples

data(dataSec)

## mark data with 1-min epoch
mydata1m = dataCollapser(dataSec, TS = "TimeStamp", col = "counts", by = 60)

data1m = wearingMarking(dataset = mydata1m,
                       frame = 90, 
                       perMinuteCts = 1,
                       TS = "TimeStamp",
                       cts = "counts", 
                       streamFrame = NULL, 
                       allowanceFrame= 2, 
                       newcolname = "wearing")

sumVct(data1m, id="dataid")

## mark data with 1-sec epoch
## Not run:

data1s = wearingMarking(dataset = dataSec,
                       frame = 90, 
                       perMinuteCts = 60,
                       TS = "TimeStamp",
                       cts = "counts", 
                       streamFrame = NULL, 
                       allowanceFrame= 2, 
                       newcolname = "wearing",
                       getMinuteMarking = FALSE)

sumVct(data1s, id="dataid")
sumVct(data1s, id="dataid", markingString = "nw")
## End(Not run)

References

Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc. 2011 Feb;43(2):357-64.

See Also

readCountsData, sumVct

Author(s)

Leena Choi leena.choi@Vanderbilt.Edu , Cole Beck cole.beck@vumc.org , Zhouwen Liu zhouwen.liu@vumc.org , Charles E. Matthews Charles.Matthews2@nih.gov , and Maciej S. Buchowski maciej.buchowski@Vanderbilt.Edu