vignette clean-up and version/date updated

DESCRIPTION

@@ -2,7 +2,7 @@ Package: lava
 Type: Package
 Title: Latent Variable Models
 Version: 1.7.4
-Date: 2023-12-10
+Date: 2023-12-20
 Authors@R: c(person("Klaus K.", "Holst", email="klaus@holst.it", role=c("aut", "cre")),
  person("Brice", "Ozenne", role = "ctb"),
  person("Thomas", "Gerds", role = "ctb"))

NEWS.md

@@ -1,16 +1,8 @@
 # lava 1.7.4
- Development version
-
  - New methods `estimate.mlm`, `IC.mlm`, `pars.mlm`, `estimate.array`, `estimate.data.frame`
- - `Print` method for nicer print of tabular data (matrix, data.frame)
- - `merge` regex 
- - `IC` should generally return row-names (default id) as obtained from model.matrix or similar
- - TODO: 
- + Rewrite DESCRIPTION/lava-package.Rd
- + Update README to reflect current state of package
- + IF vignette
- + Update help pages for merge, estimate, IC, ...
-
+ - `Print` method for tabular data (matrix, data.frame)
+ - `merge` now supports regular expressions 
+ - `IC` returns row-names (default id) as obtained from model.matrix or similar
 
 # lava 1.7.3 
 - `parameter.estimate` method to extract matrix with estimates, standard errors, and confidence limits from and estimate object (coefmat element)

vignettes/influencefunction.Rmd

@@ -41,19 +41,6 @@ knitr::opts_chunk$set(
 \newcommand{\Pz}{P_{0}}
 \newcommand{\op}{o_{P}}
 
-# TODO
-
-- Quantile regression and Zeng&Lin procedure?
-- Diagnostics Cohen's Kappa, partial gamma example
-- method argument to estimate (Richardson vs complex)
-- Explain that the `type` argument is important (first example): it defines the
- target parameter and the influence function
-- survival example in transformation section (delta method)
-- Need to elaborate on the baseline and time argument. Add documentation to
- IC.phreg and reference to that perhaps since these arguments are not documented
- in lava and not required arguments for the IC S3method
-
-
 # Influence functions
 
 Estimators that have parametric convergence rates can often be fully
@@ -356,7 +343,7 @@ semfit <- estimate(sem, data = dw)
 estimate(semfit)
 ```
 
-## TODO Example: quantile
+## Example: quantile
 
 Let \(\beta\) denote the \(\tau\)th quantile of \(X\), with IF
 \begin{align*}
@@ -365,7 +352,7 @@ Let \(\beta\) denote the \(\tau\)th quantile of \(X\), with IF
 
 where \(f_{0}\) is the density function of \(X\).
 
-To calculate the variance estimate an estimate of the density is needed which can be
+To calculate the variance estimate, an estimate of the density is thus needed which can be
 obtained by a kernel estimate. Alternatively, 
 the resampling method of [@zenglin2008] can be applied.
 Here we use a kernel smoother (additional arguments to the `estimate` function
@@ -888,146 +875,20 @@ ea
 IC(ea) |> head()
 ```
 
-## TODO: Example: Average Treatment Effects
-
+## Example: Average Treatment Effects
 
 ```{r targeted, cache=TRUE}
 a1 <- targeted::cate(a ~ 1,
+
+
+
  data = dw,
  response_model = y1 ~ x1+w+a,
  propensity_model = a ~ x1*w
  )
 a1
 IC(a1) |> head()
 ```
-
-<!-- ```{r targeted2, cache=TRUE} -->
-<!-- targeted::cate(a ~ w, -->
-<!-- data = dw, -->
-<!-- response_model = targeted::ML(y1 ~ x1+w+a, "xgboost.binary"), -->
-<!-- propensity_model = targeted::ML(a ~ x1*w, "xgboost.binary") -->
-<!-- ) -->
-<!-- ``` -->
-
-
-
-- `targeted::cate`
-- `targeted::riskreg`
-- `targeted::alean`
-- `targeted::riskreg_surv`
-- `mets::binregTSR`
-- `mets::binregATE`
-- `mets::logitATE`
-- `mets::doubleFGR`
-- `mets::resmeanATE`
-- `mets::recregIPCW`
-- `mets::recurrentMarginalAIPCWdata_`
-- `mets::normalATE`
-- `mets::binregATEbin`
-- `mets::mediatorSurv`
-- ...
-
-## TODO: Two-stage estimation
-
-```{r stack}
-stack
-```
-
-
-### TODO: Measurement error
-
-```{r measurementerror}
-measurement.error
-```
-
-
-# TODO: Examples diagnostics and ....
-
-- diagtest
-- assoc, riskcomp Ratio, Diff, odds, OR,
-- information_assoc, information.table, independence (not exported) - information.multinomial
-- gkgamma, kappa
-
-Example [@mckenzie1996; @clarke1993]
-
-50 patients from Monash Medical Centre, Melbourne
-- *Diagnosis of depression* (DSM-III-R MDD, Dysthymia, Adjustment
- Disorder with Depressed Mood, Depression NOS)
-- *Beck Depression Inventory (BDI)* (Beck et al., 1961)
-- *General Health Questionnaire (GHQ)* (Goldberg & Williams, 1988)
- (classification, see McKenzie, 1992).
-
-| Diagnosis | BDI | GHQ | n |
-|-----------+-----+-----+----|
-| \(+\) | 1 | 1 | 7 |
-| \(-\) | 1 | 1 | 1 |
-| \(-\) | 1 | 0 | 1 |
-| \(+\) | 0 | 1 | 4 |
-| \(-\) | 0 | 1 | 2 |
-| \(+\) | 0 | 0 | 2 |
-| \(-\) | 0 | 0 | 33 |
-
-Agreement between a diagnosis of depression,
-and caseness according to the screening test?
-
-Let \(X\in\{0,1,...,k-1\}\) and \(Y\in\{0,1,...,l-1\}\) 
-be categorical s.v. with the joint probabilities \(\bm{\pi}\) and data \(\bm{n}\)
-
-\begin{align*}
-\pi_{ij} = \pr(X=i,Y=j), \qquad
-\bm{n}= 
-\begin{pmatrix}
-n_{11} & n_{12} & \cdots & n_{1l} \\
-n_{21} & n_{22} & \cdots & n_{2l} \\
-\vdots & \vdots & \ddots & \vdots \\
-n_{k1} & n_{k2} & \cdots & n_{kl}
-\end{pmatrix}
-\end{align*}
-\vspace*{-.5em}
-and association measure 
-# (twice continuous differentiable)
-\begin{align*}
-\alpha = \alpha(\bm{\pi}).
-\end{align*}
-
-We will assume that we have an asymptotically linear estimator of
-\(\bm{\pi} = \bm{\pi}(\theta)\) ,
-e.g., under multinomial distribution with likelihood
-\begin{align*}
-L(\pi_{ij}\mid n_{ij}) = A\prod_{i,j} \pi_{ij}^{n_{ij}}, \pi_{ij}\geq 0, \sum_{ij}\pi_{ij}=1
-\end{align*}
-\begin{align*}
-\widehat{\pi}_{ij} = \frac{n_{ij}}{\sum_{i,j}n_{ij}}, \quad \widehat{\alpha} = \alpha(\widehat{\bm{\pi}})
-\end{align*}
-
-
-- estimate.array
-- multinomial
-
-# TODO: Enabling new models 
-
-- IC method
-- score method
-
-```{r minimal}
-minimal <- lvm(c(y1, y2, y3) ~ x) |>
- distribution(~ id1 + id2 + id3, value = list(c(1, 2, 3, 4, 5), c(1, 1, 2, 2, 2), c(1, 2, 6, 7, 8))) |>
- distribution(~ y1 + y2 + y3, value = binomial.lvm()) |>
- sim(5, seed = 123)
-
-l1 <- glm(y1~x,family=binomial,data=minimal)
-l2 <- glm(y2~x,family=binomial,data=minimal)
-l3 <- glm(y3~x,family=binomial,data=minimal)
-
-merge(l1, l2, l3) |> iid()
-merge(l1, l2, l3, id=NULL) |> iid()
-g <- merge(l1, l2, l3, id = with(minimal, list(id1, id2, id3)))
-iid(g)
-g
-estimate(l2, id = minimal$id2) |> IC()
-ii <- IC(l2)
-```
-
 # SessionInfo
 
 ```{r sessionInfo}