Design-based inference on Bernstein type estimators for continuous populations

Area of Science:

• Spatial statistics
• Geostatistics
• Numerical analysis

Background:

• Spatial estimation is crucial for understanding variable distribution across study areas.
• Traditional methods may lack precision, especially with grid-based sampling.
• Bias is a significant factor affecting the accuracy of spatial estimates.

Purpose of the Study:

• To estimate spatial variable values using Bernstein polynomials with grid sampling.
• To evaluate the precision of these estimates within a design-based framework.
• To propose and assess a pseudo-jackknife estimator for reducing bias and improving accuracy.

Main Methods:

• Bernstein polynomial approximation for spatial data.
• Design-based framework for precision evaluation.
• Pseudo-jackknife estimation to address bias in mean squared error.
• Theoretical analysis and simulation studies for performance assessment.
• Application to a real-world soil survey dataset.

Main Results:

• Bernstein polynomials provide a method for spatial estimation with regular grid sampling.
• The pseudo-jackknife estimator demonstrates improved performance in reducing bias.
• Simulation studies validate the theoretical findings on estimator precision.
• The proposed methods are applicable to practical soil science surveys.

Conclusions:

• Bernstein polynomial estimation is a viable technique for spatial variable estimation.
• The pseudo-jackknife estimator offers a significant improvement in accuracy by mitigating bias.
• The study provides a robust framework for evaluating and enhancing spatial estimation techniques.