A synthetic dataset containing 42 domains with multiple response variables
representing various probability distributions (Gaussian, Poisson, Binomial,
Beta, Negative Binomial, and Gamma). Designed for testing and benchmarking
generalized linear and spatial small area estimation models (EBP / INLA).
It shares the 42-domain spatial structure of mys_proxmat.
Usage
data(sim_area)Format
A data frame with 42 rows and 14 variables:
- area
Integer domain identifier (1 to 42).
- x1
Continuous explanatory covariate.
- x2
Uniform explanatory covariate.
- y_gaussian
Gaussian direct estimator response with unsampled domains as NA.
- vardir
Direct sampling variance for Gaussian Fay-Herriot model.
- y_poisson
Count response (Poisson) with unsampled domains as NA.
- exposure
Expected population count / exposure offset for Poisson model.
- y_binomial
Number of successes (Binomial) with unsampled domains as NA.
- trials
Sample size / number of trials for Binomial model.
- y_beta
Continuous proportion response in (0, 1) for Beta regression.
- y_nbinomial
Overdispersed count response for Negative Binomial model.
- y_gamma
Skewed positive continuous response for Gamma regression.
- x_coord
Centroid x coordinate.
- y_coord
Centroid y coordinate.
Source
Simulated using sim_area_data based on the spatial
proximity structure of mys_proxmat.
Examples
data(sim_area)
head(sim_area)
#> area x1 x2 y_gaussian vardir y_poisson exposure y_binomial trials
#> 1 1 2.5206 0.5713 2.5100 0.13373 350 141 96 113
#> 2 2 0.9203 4.8878 -1.8859 0.11747 26 97 11 196
#> 3 3 2.1392 3.1023 1.3503 0.11125 427 330 137 212
#> 4 4 1.9153 4.6371 -0.9947 0.10081 83 216 35 233
#> 5 5 1.3334 1.0112 1.8221 0.04713 223 166 89 129
#> 6 6 -0.5161 1.0776 NA 0.09209 NA 174 NA 61
#> y_beta y_nbinomial y_gamma x_coord y_coord
#> 1 0.85038 37 2.5916 0.0354 0.2881
#> 2 0.12609 1 0.5153 -0.2697 0.0128
#> 3 0.54495 37 1.1038 0.2234 0.0135
#> 4 0.17570 6 1.1678 -0.0070 0.0953
#> 5 0.61406 16 1.5058 0.1958 0.2200
#> 6 NA NA NA -0.2878 -0.0202