Dev

DESCRIPTION

@@ -2,7 +2,7 @@ Package: jointCalib
 Type: Package
 Title: A Joint Calibration of Totals and Quantiles
 Version: 0.1.2
-Authors@R:
+Authors@R: 
   c(person(given = "Maciej",
              family = "Beręsewicz",
              role = c("aut", "cre"),
@@ -13,9 +13,10 @@ License: GPL-3
 Encoding: UTF-8
 RdMacros: mathjaxr
 LazyData: yes
-Depends:
+Depends: 
     R (>= 3.5.0)
-URL: https://github.com/ncn-foreigners/jointCalib, https://ncn-foreigners.github.io/jointCalib/
+URL: https://github.com/ncn-foreigners/jointCalib,
+       https://ncn-foreigners.github.io/jointCalib/
 BugReports: https://github.com/ncn-foreigners/jointCalib/issues
 Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.2.3
@@ -29,5 +30,5 @@ Imports:
   CBPS
 Suggests: 
     knitr,
-    rmarkdown
+    rmarkdown, tinytest
 VignetteBuilder: knitr

README.Rmd

@@ -28,8 +28,7 @@ knitr::opts_chunk$set(
 
 ## Details
 
-A small package for joint calibration of totals and quantiles (see [Beręsewicz and Szymkowiak (2023)](https://arxiv.org/abs/2308.13281) working paper for details). The
-package combines the following approaches:
+A small package for joint calibration of totals and quantiles for probability and non-probability surveys as well as causal inference based on observational data. The package combines the following approaches:
 
 -   Deville, J. C., and Särndal, C. E. (1992). [Calibration estimators
     in survey
@@ -47,6 +46,7 @@ package combines the following approaches:
     Inference](https://onlinelibrary.wiley.com/doi/10.1111/insr.12518),
     International Statistical Review 91, 165--192.
 
+
 which allows to calibrate weights to known (or estimated) totals and
 quantiles jointly. As an backend for calibration
 [sampling](https://CRAN.R-project.org/package=sampling)
@@ -70,17 +70,23 @@ Currently supports:
 
 -   calibration of quantiles,
 -   calibration of quantiles and totals,
--   calibration using standard calibration, empirical likelihood and
+-   calibration using standard calibration (i.e. Deville and Särndal, 1992), empirical likelihood and
     entropy balancing method,
--   covariate distribution entropy balancing for ATT and QTT (distributional entropy balancing; DEB),
--   covariate distribution balancing propensity score for ATE and QTE (distributional propensity score; DPS).
+-   covariate distribution entropy balancing for ATT and QTT (distributional entropy balancing; DEB) via the [ebal](https://CRAN.R-project.org/package=ebal) package,
+-   covariate distribution balancing propensity score for ATE and QTE (distributional propensity score; DPS) via the [CBPS](https://CRAN.R-project.org/package=CBPS) package.
 
 Further plans:
 
 -   generalized calibration via `sampling::gencalib`,
 -   calibration for Gini and other metrics,
 -   ...
 
+For details see:
+
+-   Beręsewicz M, and Szymkowiak, M. (2023). [A note on joint calibration estimators for 
+    totals and quantiles](https://arxiv.org/abs/2308.13281), working paper (arxiv 2308.13281).
+-   Beręsewicz M (2023). [Survey calibration for causal inference: a simple method to balance covariate distributions](https://arxiv.org/abs/2310.11969), working paper (arxiv 2310.11969).
+
 ## Funding
 
 Work on this package is supported by the the National Science Centre,

README.md

@@ -16,9 +16,9 @@ Statistics](https://awesome.re/mentioned-badge.svg)](http://www.awesomeofficials
 
 ## Details
 
-A small package for joint calibration of totals and quantiles (see
-[Beręsewicz and Szymkowiak (2023)](https://arxiv.org/abs/2308.13281)
-working paper for details). The package combines the following
+A small package for joint calibration of totals and quantiles for
+probability and non-probability surveys as well as causal inference
+based on observational data. The package combines the following
 approaches:
 
 - Deville, J. C., and Särndal, C. E. (1992). [Calibration estimators in
@@ -60,19 +60,32 @@ Currently supports:
 
 - calibration of quantiles,
 - calibration of quantiles and totals,
-- calibration using standard calibration, empirical likelihood and
-  entropy balancing method,
+- calibration using standard calibration (i.e. Deville and Särndal,
+  1992), empirical likelihood and entropy balancing method,
 - covariate distribution entropy balancing for ATT and QTT
-  (distributional entropy balancing; DEB),
+  (distributional entropy balancing; DEB) via the
+  [ebal](https://CRAN.R-project.org/package=ebal) package,
 - covariate distribution balancing propensity score for ATE and QTE
-  (distributional propensity score; DPS).
+  (distributional propensity score; DPS) via the
+  [CBPS](https://CRAN.R-project.org/package=CBPS) package.
 
 Further plans:
 
 - generalized calibration via `sampling::gencalib`,
 - calibration for Gini and other metrics,
 - …
 
+For details see:
+
+- Beręsewicz M, and Szymkowiak, M. (2023). [A note on joint calibration
+  estimators for totals and
+  quantiles](https://arxiv.org/abs/2308.13281), working paper (arxiv
+  2308.13281).
+- Beręsewicz M (2023). [Survey calibration for causal inference: a
+  simple method to balance covariate
+  distributions](https://arxiv.org/abs/2310.11969), working paper (arxiv
+  2310.11969).
+
 ## Funding
 
 Work on this package is supported by the the National Science Centre,

inst/tinytest/test_jointCalib.R

@@ -0,0 +1,4 @@
+
+# Placeholder with simple test
+expect_equal(1 + 1, 2)
+

inst/tinytest/test_joint_calib_cbps.R

@@ -44,6 +44,15 @@ expect_error(
                    data = dat)
 )
 
+## check one x for quantiles
+expect_silent(
+  result <- joint_calib_cbps(formula_means = ~ X1,
+                             formula_quantiles = ~ X1,
+                             treatment = ~ D,
+                             data = dat)
+)
+
+
 ## test variable selection
 
 expect_error(
@@ -65,3 +74,5 @@ expect_silent(
 )
 
 
+
+

tests/tinytest.R

@@ -0,0 +1,5 @@
+
+if ( requireNamespace("tinytest", quietly=TRUE) ){
+  tinytest::test_package("jointCalib")
+}
+

vignettes/b_survey.Rmd

@@ -25,7 +25,7 @@ knitr::opts_chunk$set(
 
 # Setup
 
-```{r setup}
+```{r setup, message=FALSE}
 library(jointCalib)
 library(survey)
 library(laeken)
@@ -45,14 +45,16 @@ set.seed(123)
 N <- 100
 x1 <- rbinom(n = 100, prob = 0.7, size = 1)
 x2 <- rlnorm(n = 100)*1000
-pop<-data.frame(x1, x2)
+pop <- data.frame(x1, x2)
 sample_df <- pop[sample(1:N, 20), ]
 sample_df$d <- as.integer(N/nrow(sample_df))
-colSums(sample_df[, c("x1", "x2")]*sample_df$d)
-with(sample_df, weightedMedian(x2, d))
+colSums(sample_df[, c("x1", "x2")]*sample_df$d) ## sample totals
+with(sample_df, weightedMedian(x2, d)) # sample median
+median(x2) # population median
 ```
 
 Calibrate using:
+
   + totals: `x1` and `x2`,
   + median for: `x2`.
 
@@ -67,15 +69,15 @@ res <- joint_calib(formula_totals = ~ x1 + x2,
                    method = "linear",
                    control = control_calib(interpolation = "linear"))
 
-for_example <- res$Xs[, c(3,4,2)] |> as.data.frame()
-for_example$w <- sample_df$d * res$g
-for_example
+df_example <- res$Xs[, c(3,4,2)] |> as.data.frame()
+df_example$w <- sample_df$d * res$g
+df_example
 ```
 
 Median of `x2` after calibration.
 
 ```{r}
-weightedQuantile(sample_df$x2, res$g, 0.5)
+weightedQuantile(df_example$x2, df_example$w, 0.5)
 ```