user_change_raw()
to allow
for user functions that return an
unnamed vector. Also added a pkgdown
website article for user_change_raw()
.
(0.1.8.5)Added on.exit()
such that workers
will be stopped if lavaan_rerun()
terminates abnormally.
(0.1.8.2)
Updated some default values of
approx_check()
to reflect new
features added to other functions.
(0.1.8.4)
Fixed a few Rd issues on CRAN. (0.l.8.6)
index_plot()
when
the extracted column has no names.
(0.1.8.1)lavaan
functions.
(0.1.6.1)lavaan_rerun()
. (0.1.6.2)Added index_plot()
for generating
an index plot for an arbitrary
statistic. (0.1.5.5)
Added user_change_raw()
for
computing case influence on user
statistics. (0.1.5.8)
pars_id()
. Users can now use
the internal lavaan
group labels
(e.g., ".g2"
, ".g3"
) to denote
parameters in a group. This method
is not recommended (noted in the
help page) but is included as an
option. (0.1.5.2)lavaan::lavCor()
to estimate
means and correlations for Mahalanobis
distance when missing data is present.
The package norm2
is no longer
needed nor suggested. (0.1.5.4)README.md
. (0.1.5.7)est_change()
,
est_change_raw()
, est_change_approx()
, and
est_change_raw_approx()
. (0.1.4.3, 0.1.4.16, 0.1.4.18,
0.1.4.19, 0.1.4.20)fit_measures_change()
and fit_measures_change_approx()
. (0.1.4.4, 0.1.4.16)mahalanobis_rerun()
and mahalanobis_predictors()
(0.1.4.5-0.1.4.6, 0.1.4.16, 0.1.4.18)influence_stat()
.
(0.1.4.7)":="
to pars_id()
. (0.1.4.13)pkgdown
site. (0.1.4.1)approx_check()
will check if the model has at least one
equality constraint. (0.1.4.2)pars_id()
. (0.1.4.12)pgkdown
websites. (0.1.4.15)print.lavaan_rerun()
, keep the format of lavaan
warnings. (0.1.4.17)fit_measures_change_approx()
only supports
selected fit measures. (0.1.4.21)cutoff_change
and largest_change
to
est_change_gcd_plot()
. (0.1.4.22)approx_check()
to check whether the input object
is supported by the approximate method. (0.1.1.2)est_change_plot()
and est_change_gcd_plot()
,
diagnostic plots for casewise influence on
parameter estimates. (0.1.1.3)pars_id_to_lorg()
. (0.1.4)skip_all_checks
to lavaan_rerun()
, allowing users
to experiment lavaan_rerun()
and other functions on
models not officially supported.est_change()
and est_change_raw()
to support
the use of operators (e.g., ~
, =~
) to select parameters.est_change()
, est_change_raw()
and
est_change_approx()
to support models with labelled
parameters. (0.1.0.9005)pars_id()
and pars_id_to_lorg()
for converting
parameter specification to identification numbers (
positions in the vector of coefficients or row numbers
in the parameter tables). (0.1.0.9006)est_change_*
functions to use pars_id()
and pars_id_to_lorg()
. (0.1.0.9007)lavaan_rerun()
to use lavaan::lavaan()
instead of update()
as the default way to rerun. (0.1.0.9008).influence_stat()
and the plot functions to support
the approximate approach. (0.1.0.9012)Added more vignettes.
lavaan_rerun()
can accept an output with inadmissible
estimates. Disabled by default. Can be enabled by
setting allow_inadmissible
to TRUE
.
Added a print method for the lavaan_rerun()
class.
Added mahalanobis_predictors()
to compute the
Mahalanobis distance using only the observed predictors.
Both mahalanobis_predictors()
and
mahalanobis_rerun()
support datasets with missing data.
lavaan_rerun()
can specify cases to
exclude and rerun by specifying the case IDs or
selecting cases based on Mahalanobis distance on
all observed variables or on residuals of observed
variables in a path model.
Used lavaan::update()
in lavaan_rerun. This is more
reliable than recreating the call.
Added implied_scores()
. It supports only single-group
path analysis models for now.