Create a combined mts from any number of mts
objects or from a list of mts objects. The resulting mts
object with contain all deviceDeploymentIDs found in any incoming
mts and will have a regular time axis covering the the entire range
of incoming data.
If incoming time ranges are non-contiguous, the resulting mts will
have gaps filled with NA values.
An error is generated if the incoming mts objects have
non-identical metadata for the same deviceDeploymentID unless
replaceMeta = TRUE.
mts_combine(
...,
replaceMeta = FALSE,
overlapStrategy = c("replace all", "replace na")
)Any number of valid mts objects.
Logical specifying whether to allow replacement of
metadata associated with deviceDeploymentIDs.
Strategy to use when data found in time series overlaps.
An mts time series object containing all time series found
in the incoming mts objects.
(A list with meta and data dataframes.)
Data for any deviceDeploymentIDs shared among mts
objects are combined with a "later is better" sensibility where any
data overlaps exist. To handle this, incoming mts objects are first
split into "shared" and "unshared" parts.
Any "shared" parts are ordered based on the
time stamp of their last record. Then dplyr::distinct() is used to
remove records with duplicate datetime fields.
With overlapStrategy = "replace all", any data records found
in "later" mts objects are preferentially retained before the "shared"
data are finally reordered by ascending datetime.
With overlapStrategy = "replace missing", only missing values in "earlier"
mts objects are replaced with data records from "later" time series.
The final step is combining the "shared" and "unshared" parts and placing them on a uniform time axis.
library(MazamaTimeSeries)
ids1 <- example_mts$meta$deviceDeploymentID[1:5]
ids2 <- example_mts$meta$deviceDeploymentID[4:6]
ids3 <- example_mts$meta$deviceDeploymentID[8:10]
mts1 <-
example_mts %>%
mts_filterMeta(deviceDeploymentID %in% ids1) %>%
mts_filterDate(20190701, 20190703)
mts2 <-
example_mts %>%
mts_filterMeta(deviceDeploymentID %in% ids2) %>%
mts_filterDate(20190704, 20190706)
mts3 <-
example_mts %>%
mts_filterMeta(deviceDeploymentID %in% ids3) %>%
mts_filterDate(20190705, 20190708)
mts <- mts_combine(mts1, mts2, mts3)
# Should have 1:6 + 8:10 = 9 meta records and the full date range
nrow(mts$meta)
#> [1] 9
range(mts$data$datetime)
#> [1] "2019-07-01 07:00:00 UTC" "2019-07-08 06:00:00 UTC"