eg1pop_dat: Horseshore Domain Simulation Data

In funstatpackages/GgAM: Generalized geoAdditive Models

Description

This dataset includes the simulated data from horseshore domain with three different triangulations and three different correlation structure of the error. Useful for simulation in geospacial models.

Usage

data("eg1pop_dat")

Format

eg1pop_dat is a list with 9 data frames: rho03, rho05 and rho07 are the population datasets; V1, V2, V3 are the verticies datasets corresponding to different triangulations; Ts are corresponding indices of vertices in Vs.

rho03, rho05 and rho07:

Y

Response variable.

m

The true values for nonlinear part.

x1

The location parameter (longitude).

x2

The location parameter (latitude).

z1, z2, z3, z4,z5, z6, z7, z8

The parametric covariates.

Details

The finite area test function based on one is proposed by Tim Ramsay (2002). To study the performance of variable selection at different correlation levels, we generate the covariates as follows: Z _ { i 1 } = - \frac { 2 } { 3 } ≤ft{ \arctan π ≤ft( ρ \frac { X _ { i 1 } } { X _ { i 2 } } + ( 1 - ρ ) U _ { i } \right) \right}, Z _ { i 3 } = \cos π ≤ft( ρ \frac { X _ { i 1 } } { X _ { i 2 } } + ( 1 - ρ ) U _ { i } \right),Z _ { i j } \sim Uniform ( - 1,1 ), j = 2, 4,...,8, U _ { i } \sim { Uniform } ( - 1,1 ). In particular, we consider the following three cases: (i) low correlation (ρ = 0.3); (ii) medium correlation (ρ = 0.5); and (iii) high correlation (ρ = 0.7).

References

Wang et al. (2018)

Examples

## see ?plbpsm help files

funstatpackages/GgAM documentation built on Nov. 4, 2019, 12:59 p.m.