version update to 0.3.3

DESCRIPTION

@@ -2,7 +2,7 @@ Package: DMRnet
 Type: Package
 Title: Delete or Merge Regressors Algorithms for Linear and Logistic
         Model Selection and High-Dimensional Data
-Version: 0.3.2.9005
+Version: 0.3.3
 Authors@R: c(person("Agnieszka", "Prochenka-Sołtys", email = "[email protected]", role = c("aut"), comment = "previous maintainer for versions <= 0.2.0"),
    person("Piotr", "Pokarowski", role = c("aut")),
    person("Szymon", "Nowakowski", email = "[email protected]", role = c("aut", "cre"), comment = c(ORCID = "0000-0002-1939-9512")))

NEWS.md

@@ -1,26 +1,14 @@
 
-# DMRnet 0.3.2.9005
+# DMRnet 0.3.3
 
 - Updating documentation to reflect that for inference the matrix must be provided without the intercept column
-- Fixing incorrect inference for columns out of first-factors-then-numerics sequence
-- Refactoring the code to distribute the no-models fix from gaussian `DMRnet` into binomial family `DMRnet` and both `SOSnet`s and `GLAMER`s
-
-# DMRnet 0.3.2.9004
-
+- Fixing incorrect inference for columns not in first-factors-then-numerics sequence
 - Candidate fix for warnings related to ([issue #33](https://github.com/SzymonNowakowski/DMRnet/issues/33))
-
-# DMRnet 0.3.2.9003
-
 - Setting a default `nlambda` value to 100 in `cv.DMRnet()` ([issue #41](https://github.com/SzymonNowakowski/DMRnet/issues/41))
-
-# DMRnet 0.3.2.9002
-
 - Fixing problems when few or no models are available in cross validation
-- Fixing a few other minor long-standing issues in GIC-indexed cross validation
 - Fixing problems when few or no models are available after `grpreg` ([issue #39](https://github.com/SzymonNowakowski/DMRnet/issues/39))
-
-# DMRnet 0.3.2.9001
-
+- 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)
 

R/SOSnet4glm.R

@@ -24,7 +24,7 @@ SOSnet4glm <- function(X, y, o, nlambda, lam, interc, maxp, lambda){
 ###############LASSO#####################
           mL <- glmnet::glmnet(Xg, y, alpha = 1, intercept = interc, family = "binomial", nlambda = nlambda, lambda = user.lambda)
 ########################################
-          bb <- postlasso_common(mL$lambda, n/2, coef(mL))
+          bb <- postlasso_common(mL$lambda, n/2, glmnet::coef.glmnet(mL))
           #the calculations were done for mL$beta in v. prior to 0.3.2.9002
           #now, instead of mL$beta (no intercept) I pass coef(mL) which include Intercept. It helps when checks on dfy variable are performed inside
           #also, I pass n/2 instead of n to make it eliminate models with too many variables as in original AP's code

R/SOSnet4lm.R

@@ -24,7 +24,7 @@ SOSnet4lm <- function(X, y, o, nlambda, interc, maxp, lambda){
 ###############LASSO#####################
           mL <- glmnet::glmnet(Xg, y, alpha = 1, intercept = interc, family = "gaussian", nlambda = nlambda, lambda = user.lambda)
 ########################################
-          bb <- postlasso_common(mL$lambda, n, coef(mL))
+          bb <- postlasso_common(mL$lambda, n, glmnet::coef.glmnet(mL))
           #the calculations were done for mL$beta in v. prior to 0.3.2.9002
           #now, instead of mL$beta (no intercept) I pass coef(mL) which include Intercept. It helps when checks on dfy variable are performed inside