Adds support to user defined TWDTW weight function
Drop support to parallel processing
Adds a minimalist function called twdtwReduceTime that is 3x faster than twdtwApply. This function can be used for high level parallel processing implemented by users
Fixes error in to - from : non-numeric argument to binary operator in “twdtwAssess”
Fixes bug in .twdtw fundtion
Adds function for fast time series classification “twdtw_reduce_time” (~3x faster than twdtwApply)
Adds dtwSat paper published on Journal of Statistical Software
Fixing bugs
Fix error in plotAccuracy
Generalizes twdwAssess to cases with only one map
Fixes error in getTimeSeries due to time series with only one no observation
New features
Include the function twdtwApplyParallel for TWDTW parallel processing using the package snow
Include writeRaster for twdtwRaster class
Improve tests and documentation
Improve memory usage of twdtwApply
Improve memory usage and speed of twdtwClassify
Auto recognition of the argument “doy” to avoid naming the argument “doy = doy”
Fixing bugs
Fix bug in twdtwAssess for class twdtwMatches
Fix bug in twdtwRaster
New features
Register TWDTW as a distance function into package proxy
Fixing bugs
Fix typos in plot labels
New features
New accuracy metrics (twdtwAssess) for classified map, including User’s and Producer’s accuracy, and area uncertainty.
Include methods for accuracy visualization (plot and LaTeX tables)
Update data set names
Rename the data sets in ordes to avoid future overwriting of functions and data sets. “example_ts” replaced with “MOD13Q1.ts”. Tthe data sets are now called:
MOD13Q1.MT.yearly.patterns Data: patterns time series MOD13Q1.patterns.list Data: patterns time series MOD13Q1.ts Data: An example of satellite time series MOD13Q1.ts.labels Data: Labels of the satellite time series in MOD13Q1.ts MOD13Q1.ts.list
Fixing bugs
Fix bug in twdtwApply wrong sign in ‘by’ argument
Fix bug in time index for twdtwApply-twdtwRaster
Include Fortran optimization
This version includes functions written in Fortran.
Obsolete features
The S4 class ‘twdtw’ no longer exists.
New features
New S4 classes: twdtwTimeSeries, twdtwMatches, and twdtwRaster.
plot methods for twdtwRaster object: ‘maps’, ‘area’, ‘changes’, and ‘distance’.
plot methods for twdtwTimeSeries objects: ‘’patterns’’ and ‘’timeseries’’.
plot methods for twdtwMatches objects: ‘’paths’‘,’‘matches’‘,’‘alignments’‘,’‘classification’‘,’‘cost’‘,’‘patterns’‘, and’‘timeseries’’.
createPattern function to create temporal patterns based on set of time series.
getTimeSeries extract time series from raster objects.
twdtwApply apply the TWDTW analysis for raster and time series objects.
New features
‘normalizeQuery’ new normalization feature for TWDTW
‘template.list’ new dataset. List of template time series
arguments ‘from’ and ‘to’ in ‘classifyIntervals’ updated to include ‘character’ or ‘Dates’ in in the format ‘yyyy-mm-dd’
Align query and template by name if names not null in ‘twdtw’ function
deprecated features
argument ‘x’ from function ‘waveletSmoothing’ is deprecated and is scheduled to be removed in the next version. Please use ‘timeseries’ instead.
argument ‘template’ from functions ‘twdtw’ and ‘mtwdtw’ is deprecated and is scheduled to be removed in the next version. Please use ‘timeseries’ instead.
argument ‘normalized’ is deprecated and is scheduled to be removed in the next version from all methods
‘createTimeSequence’ is deprecated. Use ‘getModisTimeSequence’ instead.
Fix function name. ‘classfyIntervals’ is deprecated. Use ‘classifyIntervals’ instead.
Fixing bugs
Fix plot intervals in plotClassify
replace range(x) for range(x, na.rm=TRUE) in all methods
Bug fixed in cost matrix indexing