Provide new mechanisms for variable interaction in the spatial logistic map (#772).
Introduce configurable distance metrics for cross mapping (#761).
Enable alternative styles of spatial cross-sectional embedding (#755).
Safeguard transient removal logic in spatial logistic map to prevent index errors (#744).
Adjust the default E range in the
simplex generic to 2:10 to support more robust
reconstruction of state spaces (#739).
Introduce multithreading in distance-related computations where applicable to improve runtime efficiency (#718).
Display p-value annotation for maximum library size in cross mapping visualization legend (#710).
Prevent unreliable predictions and potentially alter results when NaNs are present in cross mapping (#747).
Rename coordinate columns in population density case csv to
lon and lat (#741).
Replace logical vectors with integer index vectors for
lib and pred in simplex and
s-mapping forecasting sources (#632).
Prevent duplicate generic registrations and simplify method definitions (#627).
Unify parameter descriptions to lowercase (#570).
Introduce confidence interval ribbon support in plot method for
gccm results (#550).
Refine randomization strategy in spatial causality test (#643).
Symbolization C++ functions now compute medians from
lib subset only (#599).
Users must now use detrend instead of
trend.rm (#559).
Enable column parameter in simplex()
and smap() and rename columns parameter to
column in multiview() (#565).
gccm results where legend
labels did not correctly match the corresponding line colors
(#552).Enact fnn R API support for false nearest neighbours
method (#512).
Integrate R API and vignette for geographical cross mapping cardinality (#455).
Provide R-level API and vignette for spatial causality test (#403).
Enable custom legend texts and colors in plot method for
gccm results (#535).
Reduce computational load in vignettes (#476).
Document overall structure and usage of spEDM in a dedicated vignette (#415).
Create SSR vignette for spatial cross-sectional data
state-space reconstruction (#412).
Include references for algorithms in spEDM
(#367).
Use non-NA spatial units for lib/pred
by default (#499).
Refine internal case data (#348).
Patch memory error caused by mismatch between C++ (0-based) and R (1-based) indexing (#480).
Fix error from non-matrix input in grid-type handling due to R matrix slicing (#474).
Enable parallel.level parameter to specify parallel
granularity in gccm R API (#310).
Implement the multiview function for multiview
embedding forecasting (MVE) method (#221).
Integrate lib parameter in gccm R API
for library units selection (#278).
Set the default k to E+2 in the
gccm R API (#261).
Eliminate redundant computations at the source C++ code level (#233).
Add trend.rm option in the R API for
embedded, simplex, and smap
methods to align with gccm behavior (#191).
Refactor indexing of lag values and embedding vector generation for spatial lattice (#186,#184) and grid data (#183,#181).
Default plotting method places the legend in the top-left corner of the plot now (#325).
Refine simplex & smap output on the
R side (#263).
embedded, simplex,
smap when input data contains only one attribute column
(#246).sdsfun package
(#159).tau parameter in the C++
source code and update the R side API (#154).Implement the smap function to enable the selection
of the optimal theta parameter (#128).
Add simplex function to support selecting the
optimal embedding dimension for variables (#98).
Provide an R-level API for generating embeddings (#97).
Now bidirectional mapping in the gccm result uses a
full join structure when organized on the R side
(#118).
Support for calculating unidirectional mappings in the
gccm function (#117).
Relax gccm C++ source code libsizes
minimum value constraint of E+2 (#109).
Provide a complete GCCM workflow for spatial lattice
and grid data in the gccm vignette (#100).
Support testing causal links in GCCM with different
E and k for cause and effect variables
(#96).
Add thread settings for gccm (#94).
Add S-maps cross-prediction support to
gccm (#81).
Resolve r crash caused by invalid E #90 and k
#89 parameter
settings in gccm.
Fix incorrect Pearson correlation calculation in C++
code when input contains NA (#83).
Encapsulate the gccm function using the S4 class
(#72).
Add options for tau, k, and
progressbar in gccm (#69).
Add print and plot s3 methods for
gccm result (#64).
gccm function returns empty
results when input grid data contains NA values (#61).GCCM method for spatial lattice and
grid data using modern C++.