64th ISI World Statistics Congress

64th ISI World Statistics Congress

Inference on missing locations in Geostatistics under Preferential Sampling

Author

GF
Gustavo Ferreira

Co-author

  • D
    Dani Gamerman

Conference

64th ISI World Statistics Congress

Format: CPS Abstract

Keywords: missing-values, sampling, stochastic

Session: CPS 17 - Statistical inference

Monday 17 July 4 p.m. - 5:25 p.m. (Canada/Eastern)

Abstract

This paper deals with the inverse problem in Geostatistics in situations where the researcher performs inference about missing locations under a specific type of informative sampling design. Based on the traditional Geostatistical model, one can found in the literature methods for drawing samples from the resulting predictive distribution of the missing locations considering that the underlying stochastic process of interest is independent of the sampling design, i.e., assuming the sampling design is not preferential. However, in cases where the sampling design is preferential it is necessary to specify a joint distribution for both the stochastic process and the sampling design and a common choice is to assume log-Gaussian Cox processes and stationary Gaussian process to deal with this situation. Unfortunately, under preferential sampling, deriving a straightforward way to obtain the predictive distribution of the missing locations is a complex task. In this paper we present a methodology to obtain the predictive distribution of missing locations under preferential sampling.

Figures/Tables

dados_max

dados_min

max_NP

max_PS

min_NP

min_PS