Mention rounding #24
Comments
|
@muschellij2 Thanks for mentioning this difference in the |
|
I think the results should be rounded to 3 decimals to replicate the output
from actilife.
…On Tue, Oct 3, 2023 at 5:51 PM Brian C. Helsel ***@***.***> wrote:
@muschellij2 <https://github.com/muschellij2> Thanks for mentioning this
difference in the read.gt3x and pygt3x packages. It seems like this might
be something that needs to be fixed with the read.gt3x R package since it
is returning rounded decimal points. Any thoughts as to what part of the
code is doing the rounding or any potential solutions?
—
Reply to this email directly, view it on GitHub
<#24 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAIGPLQ25SSZJOUQHCKAFYLX5SCH5AVCNFSM6AAAAAA4LXTPVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONBVG44DGNBUG4>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
From what I've found, the library(agcounts)
library(dplyr)
#>
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#>
#> filter, lag
#> The following objects are masked from 'package:base':
#>
#> intersect, setdiff, setequal, union
path <- system.file("extdata/example.gt3x", package = "agcounts")
agd_path <- system.file("extdata/example5sec.agd", package = "agcounts")
agd <- agcounts:::.read_agd(agd_path)$data
pygt3x <-
agcounts::agread(path, parser = "pygt3x") %>%
agcounts::calculate_counts(epoch = 5)
pygt3x_round <-
agcounts::agread(path, parser = "pygt3x") %>%
dplyr::mutate_at(.vars = 2:4, .funs = round, 3) %>%
agcounts::calculate_counts(epoch = 5)
pygt3x_ncar_round <-
agcounts::agread(path, parser = "pygt3x") %>%
dplyr::mutate_at(.vars = 2:4, .funs = ncar::Round, 3) %>%
agcounts::calculate_counts(epoch = 5)
colSums(abs(pygt3x[, 2:5] - agd[, 2:5]))
#> Axis1 Axis2 Axis3 Vector.Magnitude
#> 0 0 0 0
colSums(abs(pygt3x_round[, 2:5] - agd[, 2:5]))
#> Axis1 Axis2 Axis3 Vector.Magnitude
#> 7 4 13 9
colSums(abs(pygt3x_ncar_round[, 2:5] - agd[, 2:5]))
#> Axis1 Axis2 Axis3 Vector.Magnitude
#> 7 4 16 11
read.gt3x <-
agcounts::agread(path, parser = "read.gt3x") %>%
agcounts::calculate_counts(epoch = 5)
colSums(abs(pygt3x_ncar_round[, 2:5] - read.gt3x[, 2:5]))
#> Axis1 Axis2 Axis3 Vector.Magnitude
#> 0 0 0 0Created on 2023-10-04 with reprex v2.0.2 |
|
Actigraph specification of GT3X format specifically says to round ACTIVITY packets: https://github.com/actigraph/GT3X-File-Format/blob/main/LogRecords/Activity2.md#scaling-activity-values and https://github.com/actigraph/GT3X-File-Format/blob/main/LogRecords/Activity.md#scaling-activity-values |
|
The documentation doesn't change the fact that the |
|
That’s fair, but many will likely use base::round and get slightly wrong
results. Either way, is this noted in the docs?
…On Wed, Oct 4, 2023 at 9:50 AM Brian C. Helsel ***@***.***> wrote:
The documentation doesn't change the fact that the pygt3x parser is
getting a more consistent output compared to Actilife and is the method
that ActiGraph recommended to read in the .gt3x raw files when their
agcounts Python package was released. If users want a value rounded to 3
decimal points then it seems like the read.gt3x parser would be the best
option since it rounds similar to Ncar::Round. Users could also round the
values after reading them into R and before calculating the activity counts
which yields the same output as the read.gt3x parser.
—
Reply to this email directly, view it on GitHub
<#24 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAIGPLSM27NGGNLRXPGS4KDX5VSSPAVCNFSM6AAAAAA4LXTPVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONBWHEZDCMRUGQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
We list the preferred method as |
As per the GT3X documentation, rounding half away from zero is the norm for rounding GT3X data as the last step (https://github.com/actigraph/GT3X-File-Format/blob/d4aa795a3d06cc8b22b4e3df645b597cdeee6d92/LogRecords/Activity.md?plain=1#L226). It should be noted somewhere that the normal
roundin R does not do this, but thencar::Roundfunction will (as well as many others).This is important because you can get some slight differences from the readers:
Created on 2023-09-05 with reprex v2.0.2
Session info
The text was updated successfully, but these errors were encountered: