Changes in version 2.4.0
-
ADD: new function (
plot_clusters) to plot clusters of individualdynEGAresults -
ADD: TEFI optimization for
dynEGAfunctions -
ADD:
dynamic.network.compareto compare dynamic networks (allows for paired and non-paired tests) -
ADD:
network.fitfor traditional fit statistics (only accurate for continuous data) -
ADD:
known.graphto refit without regularization or allow a graph to be estimated with weights when it is known -
CHANGE:
network.nonconvexchanged tonetwork.regularizationfor more generic implementations -
UPDATE: zero variance time series handling in
dynEGAfunctions -
UPDATE: better mechanism for
simEGM -
UPDATE: better optimizations for
EGM.optimizations -
REMOVE:
EGMhas been removed (for now) as its under revision -
REMOVE:
network.nonconvexhas been removed (for now) as its under revision -
INTERNAL: regularization penalties and derivatives are added but not used (yet)
Changes in version 2.3.1
-
FIX: documentation for
itemDiagnostics -
FIX:
$keepoutput fromitemDiagnosticsrather than$suggested -
UPDATE: added argument,
ordered, to either outputnet.loadsin the initial variable order or descending order for each community -
UPDATE: beta-min condition is used to supply a model-implied network for
simEGM -
UPDATE: missing data handling in
gllaand subsequentlydynEGAfunctions
Changes in version 2.3.0
-
ADD:
itemDiagnosticsfor automated detection of potential instability issues such as local dependnece, minor dimensions, multidimensionality, and low loadings -
FIX: default
itemStabilityplot switched back to empirical dimensions only -
UPDATE:
simEGMhas a new (faster) data generating mechanism -
UPDATE:
EGMoptimizations for log-likelihood, AIC, and BIC were added (including proper gradients)