Package 'saePseudo'

Small Area Estimation using Averaging Pseudo Area Level Model

Description

Provides function for small area estimation at area level using averaging pseudo area level model for variables of interest. A dataset produced by data generation are also provided. This package estimates small areas at the village level and then aggregates them to the sub-district, region, and provincial levels.

Usage

avgPseudo(prov, reg, sub, vill, y, x, var, N, method = "REML")

Arguments

prov	Vector containing information of province
reg	Vector containing information of region
sub	Vector containing information of subdistrict
vill	Vector containing information of village
y	Direct estimation for each area
x	Auxiliary variable for each area
var	Sampling variances of direct estimators for each domain
N	Number of population in each area
method	Method used to fit the Fay-Herriot model, which can be either "ML", "REML" or "FH" methods. Default is "REML" method

Value

This function returns a list of the following objects:

Est_Area3	A dataframe with the values of Small Area Estimation with averaging pseudo area level model for sub-district level
Est_Area2	A dataframe with the values of Small Area Estimation with averaging pseudo area level model for region level
Est_Area1	A dataframe with the values of Small Area Estimation with averaging pseudo area level model for provincial level

Examples

# Load Dataset
data(dataVill)
saeAVG.Pseudo <- avgPseudo(prov = dataVill$Area1, reg = dataVill$Area2, sub = dataVill$Area3,
                 vill = dataVill$Area4, y = dataVill$ydir_area4, x = dataVill$X1,
                 var = dataVill$vardir_area4, N = dataVill$N, method="REML")

# Result
saeAVG.Pseudo$Est_Area3
saeAVG.Pseudo$Est_Area2
saeAVG.Pseudo$Est_Area1

---

Sample Data for Small Area Estimation using Averaging Pseudo Area Level Model

Description

Dataset to simulate Small Area Estimation using Averaging Pseudo Area Level Model This data is generated by these following steps:

1. Generate population data consisting Area1 (province), Area2 (region), Area3 (sub-district), Area4 (village), and Unit. The auxiliary variabels are generated by Uniform distribution with $(x1 ~ U(40, 100))$ and Normal distribution with $(x2 ~ N(70, 5))$. The coefficient parameters are set as $\beta_{0} = 0.5$, $\beta_{1} = 0.2$, and $\beta_{2} = 0.2$

2. Calculate $y_{k} = \beta_{0}+\beta_{1}*x1_{k}+\beta_{2}*x_2{k}$

3. Generate number of sample with simple random sampling with replacement

4. Calculate $ydir_area4 = \frac{\Sigma{y_{k}}}{n}$, $vardir_area4 = \frac{\Sigma{(y_{k}-\frac{\Sigma{y_{k}}}{n})^2}}{n-1}$, and auxiliary variable $X1 = \frac{\Sigma{x1_{k}}}{n}$

5. Calculate N (number of population) based on the initial population generated for each Area4 selected as a sample

6. Area1, Area2, Area3, Area4, ydir_area4, vardir_area4, X1, and N are combined in a dataframe called dataVill.

Usage

dataVill

Format

A data frame with 83 observations on the following 8 variables:

Area1

Province

Area2

Region

Area3

Sub-district

Area4

Village

ydir_area4

Direct Estimation of y

vardir_area4

Sampling variance of y

X1

Auxiliary variable

N

Number of population in area4

---