-
Notifications
You must be signed in to change notification settings - Fork 1
/
README.Rmd
118 lines (84 loc) · 3.42 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r include=FALSE }
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "50%"
)
```
<!-- badges: start -->
[](https://travis-ci.com/kkholst/targeted)
[](https://app.codecov.io/github/kkholst/targeted?branch=main)
[](
https://opensource.org/license/apache-2-0)
[](https://cranlogs.r-pkg.org/downloads/total/last-month/targeted)
<!-- badges: end -->
# Targeted Inference in R: targeted <a href="https://kkholst.github.io/targeted"><img src="man/figures/logo.svg" align="right" height="75" /></a>
<!-- # Targeted Inference in R: targeted <a href="https://targetlib.org/r/"><img src="man/figures/logo.svg" align="right" height="75" /></a> -->
Various methods for targeted and semiparametric inference including augmented
inverse probability weighted (AIPW) estimators for missing data and causal
inference (Bang and Robins (2005) <doi:10.1111/j.1541-0420.2005.00377.x>),
variable importance and conditional average treatment effects (CATE) (van der
Laan (2006) <doi:10.2202/1557-4679.1008>), estimators for risk differences and
relative risks (Richardson et al. (2017) <doi:10.1080/01621459.2016.1192546>),
assumption lean inference for generalized linear model parameters (Vansteelandt
et al. (2022) <doi:10.1111/rssb.12504>).
## Installation
You can install the released version of targeted from [CRAN](<https://CRAN.R-project.org>) with:
```{r install, eval=FALSE }
install.packages("targeted")
```
And the development version from [GitHub](<https://github.com/kkholst/targeted>) with:
```{r eval=FALSE }
remotes::install_github("kkholst/targeted")
```
## Examples
### Targeted risk regression
```{r results='hide' }
library(targeted)
```
Simulate some data:
```{r }
m <- lvm() |>
regression(a ~ x+z) |>
regression(lp.target ~ 1) |>
regression(lp.nuisance ~ x + z) |>
distribution('a', binomial.lvm("logit")) |>
binomial.rr('y', 'a', 'lp.target', 'lp.nuisance')
par <- c('a'=-2, 'lp.target'=1, 'lp.nuisance'=-1, 'lp.nuisance~x'=2)
d <- lava::sim(m, n=1e4, seed=1, p=par) |>
subset(select=c('y', 'a','x','z'))
head(d)
```
```{r }
fit <- riskreg(y ~ a, nuisance=~x+z, data=d, type="rr")
fit
```
Here the same design matrix is used for both the propensity model and
the nuisance parameter (odds-product) model
```{r }
summary(fit)
```
Double-robustness illustrated by using a wrong propensity
model but a correct nuisance paramter (odds-product) model:
```{r }
riskreg(y ~ a, nuisance=~x+z, propensity=~z, data=d, type="rr")
```
Or vice-versa
```{r }
riskreg(y ~ a, nuisance=~z, propensity=~x+z, data=d, type="rr")
```
Whereas the MLE yields a biased estimate of the relative risk:
```{r }
fit_mle <- with(d, riskreg_mle(y, a, x1=model.matrix(~1,d), x2=model.matrix(~z, d)))
estimate(fit_mle, 1)
```
To obtain an estimate of the risk-difference (here wrong model) we simply chance the `type` argument
```{r }
riskreg(y ~ a, nuisance=~x+z, data=d, type="rd")
```
Interactions with the exposure can be examined with the `target` argument
```{r }
riskreg(y ~ a, target=a~x, nuisance=~x+z, data=d, type="rr")
```