DPI 2025.11
- Improved
DPI_dag()
. - Fixed
DPI_curve()
for wrong (reverse) direction of DPI caused by the change of parameter order ofx
andy
in version 2025.10. - Fixed a bug caused by
dpi
parameter-object name conflict (internally) when savingDPI()
results into afile
.
DPI 2025.10
CRAN release: 2025-10-16
This version contains breaking changes to function names and visualization methods.
- Added
DPI_dag()
: Directed acyclic graphs (DAGs) via DPI exploratory analysis (causal discovery) for all significant partial correlations. - Added
bonf
andpseudoBF
parameters toDPI()
,DPI_curve()
, andDPI_dag()
.-
bonf
: Bonferroni correction to control for false positive rates among multiple pairwise DPI tests. -
pseudoBF
: Use normalized pseudo Bayes Factorssigmoid(log(PseudoBF10))
as the Significance score (0~1). Pseudo Bayes Factors are computed using the transformation rules proposed by Wagenmakers (2022) https://doi.org/10.31234/osf.io/egydq.
-
- Added S3 methods
plot.cor.net()
,plot.bns.dag()
, andplot.dpi.dag()
that can transformqgraph
base-plot objects intoggplot
objects for more stable and flexible visualization. - Added
p_to_bf()
: Convert p values to pseudo Bayes Factors (). - Renamed
cor_network()
tocor_net()
,dag_network()
toBNs_dag()
, andmatrix_cor()
tocor_matrix()
. - Fixed
cor_net()
to return the exactly correct p values of (partial) correlation coefficients. - Improved output information in console and plot.
DPI 2025.9
CRAN release: 2025-09-20
This version contains breaking changes to both algorithm and functionality.
- Refined
DPI()
algorithm to limit and also simplified its output information.- In an earlier version of algorithm, the strength score was computed as . While this algorithm performs as well as the new approach (e.g., both have low false positive and false negative rates), has a major flaw that its values cannot converge to a limited range so that the final DPI values would be heavily determined by , which is not a desired attribute. In contrast, the new algorithm can make the significance score more likely to be an “on-off switch”, with values more likely approximating 0 or 1, thereby minimizing its impact on the interpretation of final DPI values.
- Renamed
data_random()
tosim_data()
with enhanced functionality that supports data simulation from a multivariate normal distribution, usingMASS::mvrnorm()
. - Added
sim_data_exp()
: Simulate experiment-like data with independent binary Xs. - Used
gc()
inDPI()
,DPI_curve()
, anddag_network()
for memory garbage collection. - Provided a better example in
dag_network()
for arranging multiple base-R-style plots usingaplot::plot_list()
.
DPI 2025.8
CRAN release: 2025-08-20
- Added
dag_network()
: Directed acyclic graphs (DAGs) via causal Bayesian networks (BNs). - Improved
cor_network()
: Correlation and partial correlation networks. - Moved help pages of all S3 method functions to
S3method.dpi
andS3method.network
and made them as internal topics.
DPI 2025.6
CRAN release: 2025-06-18
- CRAN package publication.
- Initial public release on GitHub.
- Developed core functions and package logo.