NEWS
DMRnet 0.4.0.9000
- Starting a new development version
DMRnet 0.4.0 (2023-08-07)
- Variable selection algorithm
var_sel
added
- GLAMER implemented as a net version respective to its $\tau$ parameter (issue #57)
- Features package load message with citation info
- Fixing a problem with zero-length arrow in CV plots resulting from sd=0 (problem still remaining in issue #39 after the
0.3.2.9002
fix)
- Citation update
- Hierarchical clustering based on segmentation of 1D points with
hclust1d
package for PDMR and GLAMER
DMRnet 0.3.4 (2023-01-31)
DMRnet 0.3.3 (2022-11-15)
- Updating documentation to reflect that for inference the matrix must be provided without the intercept column
- Fixing incorrect inference for columns not in first-factors-then-numerics sequence
- Candidate fix for warnings related to issue #33
- Setting a default
nlambda
value to 100 in cv.DMRnet()
(issue #41)
- Fixing problems when few or no models are available in cross validation
- Fixing problems when few or no models are available after
grpreg
(issue #39)
- Refactoring the code to distribute the no-model fixes from gaussian
DMRnet
into binomial family DMRnet
and both SOSnet
s and GLAMER
s
- Fixing a few other minor long-standing issues in GIC-indexed cross validation
- Fixing df.1se in GIC-indexed cross validation for binomial GLAMER
- Fixing incorrect loglik calculation for the first (largest) model in binomial family, for both
GLAMER
(-Inf
) and DMRnet
(incorrect)
DMRnet 0.3.2 (2022-10-07)
- Improved readability of a getting-started vignette
- Fixed a bug in model-indexed cross validation related to folds with different model sizes
- Added df.1se to GIC-indexed cross validation
- Improved CV plots with
df.1se
model
- Improved readability of README on CRAN (issue #32)
- Welcome message on package load added
DMRnet 0.3.1 (2022-07-19)
Major changes
- GLAMER (Group LAsso MERger) algorithm added
- Rebuild of the existing cross validation routine (models indexed by model dimension)
- New cross validation routine added (models indexed by GIC) and made a new default
- Handling of mismatched factor levels restructured
Stability improvements
- Adding pivots in SOSnet for gaussian families
- Adding a
try
-catch
clause to recalculate QR decomposition for pivoted columns within rank in case of failure
- Adding the regularization in gaussian family in
DMR
direct calls and calls via DMRnet
- fixing NA values resulting in attempting non-penalized regression in case of non-full rank design matrices
Other improvements
- Fixing how the parameters get passed in
plot
family of functions
- Returning only non-zero coefficients in
coef
- Fixing
DRMnet
and cross validation in handling data with a single two-factor column
- Fixing negative Wald statistics in binomial case
- Workaround in GLAMER for cases of
grpreg
not observing a group constraint