v0.9 major dev update

.Rbuildignore

@@ -21,3 +21,10 @@
 ^CRAN-SUBMISSION$
 ^debugging$
 ^debugger$
+^\.cursor$
+^\.discussion$
+^\.discussions$
+^GEMINI\.md$
+^README.html$
+^R/ddml_to_did\.R$
+^tests/testthat/test-ddml_to_did\.R$

.github/workflows/pkgdown.yaml

@@ -29,13 +29,10 @@ jobs:
           #r-version: '4.1.0'
           use-public-rspm: true
 
-      ## FIX??
-      #- name: Install lme4
-      #  run: Rscript -e "install.packages('lme4')"
 
       - uses: r-lib/actions/setup-r-dependencies@v2
         with:
-          extra-packages: any::pkgdown, local::., cran/lme4@1.1-35
+          extra-packages: any::pkgdown, local::.
           needs: website
 
       - name: Build site
@@ -44,7 +41,7 @@ jobs:
 
       - name: Deploy to GitHub pages 🚀
         if: github.event_name != 'pull_request'
-        uses: JamesIves/github-pages-deploy-action@v4.5.0
+        uses: JamesIves/github-pages-deploy-action@v4
         with:
           clean: false
           branch: gh-pages

.gitignore

@@ -9,3 +9,10 @@ inst/doc
 .direnv
 /debugging/
 /debugger
+.cursor/
+.discussion/
+.discussions/
+GEMINI.md
+README.html
+R/ddml_to_did.R
+tests/testthat/test-ddml_to_did.R

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: ddml
 Title: Double/Debiased Machine Learning
-Version: 0.3.1
-Date: 2025-12-10
+Version: 0.9.0
+Date: 2026-03-06
 Authors@R: c(
     person("Achim", "Ahrens", role = "aut"),
     person("Christian B", "Hansen", role = "aut"),
@@ -23,21 +23,26 @@ RoxygenNote: 7.3.3
 Depends: 
     R (>= 4.3)
 Imports: 
-    methods,
-    stats,
-    AER,
+    generics,
+    glmnet,
+    graphics,
     MASS,
     Matrix,
+    methods,
     nnls,
+    pbapply,
     quadprog,
-    glmnet,
     ranger,
+    stats,
     xgboost
 Suggests: 
-    sandwich,
+    AER,
     covr,
-    testthat (>= 3.0.0),
+    CVXR,
     knitr,
-    rmarkdown
+    parallel,
+    rmarkdown,
+    sandwich,
+    testthat (>= 3.0.0)
 Config/testthat/edition: 3
 VignetteBuilder: knitr

NAMESPACE

@@ -1,34 +1,88 @@
 # Generated by roxygen2: do not edit by hand
 
+S3method("[",summary.ddml)
+S3method("[[",ral_rep)
+S3method(as.list,ddml)
+S3method(as.list,ddml_rep)
+S3method(as.list,ral)
+S3method(as.list,ral_rep)
+S3method(coef,ral)
+S3method(coef,ral_rep)
+S3method(confint,ral)
+S3method(confint,ral_rep)
+S3method(glance,ddml)
+S3method(glance,ddml_rep)
+S3method(glance,ral)
+S3method(glance,ral_rep)
+S3method(hatvalues,ral)
+S3method(length,ral_rep)
+S3method(lincom,ddml)
+S3method(lincom,ddml_rep)
+S3method(nobs,ral)
+S3method(nobs,ral_rep)
+S3method(plot,ral)
+S3method(plot,ral_rep)
 S3method(predict,ensemble)
+S3method(predict,mdl_bigGlm)
 S3method(predict,mdl_glm)
 S3method(predict,mdl_glmnet)
 S3method(predict,mdl_ranger)
 S3method(predict,mdl_xgboost)
 S3method(predict,ols)
-S3method(print,summary.ddml_ate)
-S3method(print,summary.ddml_att)
-S3method(print,summary.ddml_fpliv)
-S3method(print,summary.ddml_late)
-S3method(print,summary.ddml_pliv)
-S3method(print,summary.ddml_plm)
-S3method(summary,ddml_ate)
-S3method(summary,ddml_att)
-S3method(summary,ddml_fpliv)
-S3method(summary,ddml_late)
-S3method(summary,ddml_pliv)
-S3method(summary,ddml_plm)
+S3method(print,ddml_diagnostics)
+S3method(print,ddml_rep)
+S3method(print,lincom)
+S3method(print,lincom_rep)
+S3method(print,ral_rep)
+S3method(print,summary.ddml)
+S3method(print,summary.ddml_rep)
+S3method(print,summary.ral)
+S3method(print,summary.ral_rep)
+S3method(summary,ddml)
+S3method(summary,ddml_rep)
+S3method(summary,ral)
+S3method(summary,ral_rep)
+S3method(tidy,ddml)
+S3method(tidy,ddml_diagnostics)
+S3method(tidy,ddml_rep)
+S3method(tidy,ral)
+S3method(tidy,ral_rep)
+S3method(vcov,ral)
+S3method(vcov,ral_rep)
 export(crosspred)
 export(crossval)
+export(ddml)
+export(ddml_apo)
 export(ddml_ate)
 export(ddml_att)
+export(ddml_attgt)
 export(ddml_fpliv)
 export(ddml_late)
 export(ddml_pliv)
 export(ddml_plm)
+export(ddml_rep)
+export(ddml_replicate)
+export(diagnostics)
+export(ensemble)
+export(ensemble_weights)
+export(glance)
+export(lincom)
+export(lincom_weights_did)
+export(mdl_bigGlm)
 export(mdl_glm)
 export(mdl_glmnet)
 export(mdl_ranger)
 export(mdl_xgboost)
 export(ols)
+export(ral)
+export(ral_rep)
 export(shortstacking)
+export(tidy)
+importFrom(generics,glance)
+importFrom(generics,tidy)
+importFrom(graphics,plot)
+importFrom(stats,coef)
+importFrom(stats,confint)
+importFrom(stats,hatvalues)
+importFrom(stats,nobs)
+importFrom(stats,vcov)

NEWS.md

@@ -1,9 +1,24 @@
-# ddml 0.3.1
-
-* Updates internals of ``ddml::mdl_xgboost()`` with new ``xgboost`` syntax.
-* Fixes ``ddml::ddml_fpliv()`` with custom weights.
-* Allows for stacking with no positive stacking weights.
-* Fixes ``ddml::mdl_glmnet`` predictions for binomial regression.
+# ddml 0.9.0
+
+* Adds `ddml_attgt()` for staggered DiD and `ddml_apo()` for average potential outcomes.
+* Adds `ddml()` constructor for custom DML estimators with user-supplied scores.
+* Adds `lincom()` and `lincom_weights_did()` for inference on linear combinations.
+* Influence-function-based inference via the `ral` class; all estimators now inherit from `ral`.
+* Adds `ddml_rep()` and `ddml_replicate()` for repeated cross-fitting with median, mean, or spectral-norm aggregation.
+* Adds `diagnostics()` for MSPE, R-squared, stacking weights, and CVC tests.
+* Adds `fitted`/`splits` pass-through to all `ddml_*()` estimators.
+* New S3 methods: `plot()`, `as.list()`, `hatvalues()`, `nobs()`, multi-ensemble `tidy()`/`glance()`.
+* Adds uniform confidence bands via multiplier bootstrap (`confint(uniform = TRUE)`).
+* Adds HC0/HC3 variance estimators, parallel computation, stratified cross-fitting, cluster-aware splitting, and input validation.
+* Adds `broom` compatibility.
+* Fixes `ddml_fpliv()` with custom weights.
+* Breaking changes:
+    - Inference internals use `$inf_func` instead of `$scores`/`$J`/`$psi_a`/`$psi_b`.
+    - Utility functions (`crosspred`, `crossval`, `ensemble`, `ensemble_weights`, `shortstacking`) no longer accept `Z`/`newZ`. Pre-concatenate instruments with covariates (e.g., `cbind(X, Z)`).
+    - `crosspred()` and `shortstacking()` drop `compute_insample_predictions` and `insample_fitted` output.
+    - `ddml_fpliv()` drops the `enforce_LIE` argument.
+    - `shortstacking()` drops `shortstack_y`.
+    - `ddml_*()` estimators drop `subsamples`, `cv_subsamples`, `subsamples_byD`, `cv_subsamples_byD`. Use the new `splits` parameter instead.
 
 # ddml 0.3.0
 

R/AE98.R

@@ -1,6 +1,6 @@
-#' Random subsample from the data of Angrist & Evans (1991).
+#' Random Subsample from the Data of Angrist & Evans (1998)
 #'
-#' @description Random subsample from the data of Angrist & Evans (1991).
+#' @description Random subsample from the data of Angrist & Evans (1998).
 #'
 #' @format A data frame with 5,000 rows and 13 variables.
 #' \describe{