3.1-0

• Minor issue fixes introduced in 3.0-10 related to handling nights without sustained inactivity bouts, handling recordings without midnight and storage of GGIR version in the part 4 output when there are no valid nights.

• Part 5 report: Add option to filter part 5 windows based on parameter includedaycrit originally used in part 2, for details see documentation of parameter includedaycrit and #1113

• Part 1: Fix bug in identifying and naming the vector magnitude column #1122

3.0-10

• Part 1:

  • Change default for parameter "minloadcrit" to 168 hours, see #1101.

  • Improve handling of failure to extract sampling frequency from gt3x file #1098

  • Fix bug in handling ad-hoc csv format data expressed in mg units #1110

  • Fix bug introduced in 3.0-6 affecting handling of time gaps, #1104

• Part 1-6: GGIR release number now stored in each milestone data file and in each
  report to ease tracing what release was used per recording and per GGIR part, see #1101.

• Part 3: Revert change to HDCZA in 3.0-7, see issue #1102

• Part 4: Fixes issue #1095 introduced in 3.0-7 that causes GGIR part 4 to stop when processing data without non-default sleep algorithms.

• Part 2 and 5:

  • qwindow functionality enhanced to also consider fractions of minutes, #1093

  • Fix minor bug in activity diary loading when date format is %Y-%m-%d, #1090

• Part 5 and 6: Correct algorithm for coefficient of variance (CoV) in g.fragmentation, #1105

3.0-9

• Part 5: Temperature (if available) added to time series output #1085.

• Part 5: Fix minor bug in merging night level sleep variables into part 5 report, this does not affect the main part 5 estimates such as time spent in intensity ranges #1086.

• Part 4: Allow handling sleeplog with only one record #1083

• General: simplify installation for typical use cases by moving GGIRread, ActCR and read.gt3x to imports.

EDIT: release accidentally had DESCRIPTION file with 3.0-8 still in it. It seems I cannot correct that now.

3.0-8

• Part 1: In the handling of externally derived epoch data, the code and algorithm for nonwear detection is now simplified to better match expected behaviour #1080.

• Part 2: Fixed issue where g.convert.part2.long() was throwing an error when attempting to process data where every day had insufficient number of valid hours #1070.

• Part 5: Fix bug introduced with 3.0-7 causing WW window to not handle well scenario of zero windows #1078.

3.0-7

• Part 1:

  • Fix bug introduced with release 3.0-6 affecting the use of external function embedding #1065

  • Need to specify rmc.firstrow.acc when working with ad-hoc csv file format now better documented and read.myacc.csv now produces an error when used directly and user forgets to specify rmc.firstrow.acc #1034.

  • Improve g.calibrate to better handle scenario when no non-movement periods are found in the entire recording #1032

• Part 3: Algorithm HDCZA simplified by replacing time series specific threshold in step 6 of the description in the 2018 paper by a constant threshold that can be set by the user. This means that
  we have a new parameter HDCZA_threshold and parameter constrain2range is now deprecated #1062.

• Part 3: Added option for HSPT.ignore.invalid = NA, which would consider invalid
  time segments as no movement for the Sleep Period Time definition.

• Part 5: Fix bug in MM = timewindow part specific to when first night(s) are not available in part 4 results. #1039.

3.0-6

• Part 2, 4, 5: Add parameter dec_reports and dec_config to ease tailoring GGIR to non-default UK/US machines #1048.

• Part 2, 4, 5: Force language to "C" (UK/US English) to avoid issues on computer configured differently #1047.

• Part 5: Behaviour parameter includedaycrit.part5 changed for values above 1, these
  are now interpreted as minimum number of valid waking hours during waking horus of a day.
  If you prefer to keep old functionality then divide your old value by 24 #1050.

• Part 1:

  • Improved readability and maintainability of the code #1027

  • Improved processing speed for Axivity .cwa, GENEActiv .bin, and Movisens files

  • Made sure that g.readaccfile() reads timestamps in the correct timezone, configtz, for all monitor types

  • Note: there will be small differences in both metalong and metashort metrics calculated by this GGIR version, compared to prior versions. This is due to small improvements in the management of timestamps, calibration coefficients, and input data block boundaries.

  • Fix handling of ad hoc csv file header in g.inspectfile() #1057

  • Improved g.calibrate to better handle scenario when no non-movement periods are found in the entire recording #1032

  • Improved documentation for the need to specify rmc.firstrow.acc when working with ad-hoc csv file format and read.myacc.csv now produces an error when used directly while user forgets to specify rmc.firstrow.acc #1034.

3.0-5

• Part 1: For externally derived epoch data in dataFormat actiwatch_csv, actiwatch_awd, actigraph_csv, and sensewear_xls the non-wear detection is no longer done based on a hard-coded 60 minute rolling window but the window length is now modifiable with the third value of argument windowsizes (in seconds) as also used for raw data. #1026
• Part 5: Fix issue with defining days specific to when timewindow is MM, recordings starts at midnight and both first and last days are incomplete #1029
• Part 5: Fix bug in functionality for Sensewear data (externally derived epoch data) #1030

3.0-4

• Part 5: Improved handling of inconsistent number of columns in part 5 milestone data #1002

• Part 3: Revised NotWorn option for argument HASPT.algo as used for count data #1011

• Visualreport: Now able to handle recordings with no valid days in combination with argument visualreport_without_invalid = TRUE, fixes bug #1016

• Part 2 and 5: Timing of LX is now expressed on scale between 12 and 36 to allow for meaningful person level summary of this value, fixes #1012

• Facilitate handling of Sensewear xls file format with externally derived epoch data #974

• Part 3: If argument sensor.location is set to hip then use longitudinal axis detected in part 2 instead of estimating it again with a different algorithm #1024

• Part 5: If argument sensor.location is set to hip then store and use longitudinal angle and possible other angles in part 5 time series instead of anglez #1024.

3.0-3

• Part 2: Fix bug where data_quality_report.csv contained incorrect filehealth[...] values whenever some of these values were supposed to be blank #1003

• Part 1: Fix bug where on machines with GMT timezone and R >= 4.3.0, for GENEActiv .bin files, the starting timestamps of M$metalong and M$metashort were truncated to midninght #1000

• Part 5: Fix bug where part5 output showed duplicate rows when argument dayborder is set to nondefault value #1006

3.0-2

• Part 2: Fix bug that caused part 2 to struggle with corrupt ActiGraph .gt3x files #972

• Part 2: Redefine horizontal axis of plots_to_check_data_quality #983

• Documentation: Expanded documentation on desiredtz, configtz, and time stamp format in part 5 time series #966

• Part 1: Now also able to handle some more variations in Actigraph csv count file format #978, and automatically aggregates to lower resolution if short epoch size is longer than observed epoch size in actigraph count csv.

• Part 5: Reverting decision to prohibit segmentDAYSPTcrit.part5 to be c(0, 0). The default remains unchanged and documentation now only emphasizes the downside of using c(0, 0). #980

• Argument documentation: Fixing series of typos (thanks to Pieter-Jan Marent for pointing them out)

• Part 5: Fix bug in recently added functionality for studying overlap between sibs and self-reported behaviours #989.

• Part 1:Enable use of temperature data in adhoc-csv format throughout GGIR #991

• Part 1: Fix bug that caused a mismatch between IDs and filenames in part 1 when movisens participant folders did not contain the acc.bin file #994.