# Nearest-Neighbor Neural Networks for Geostatistics

## TL;DR

This paper introduces the 4N process, a flexible deep learning-based spatial prediction method that outperforms traditional geostatistical models on non-Gaussian data and large forestry datasets.

## Contribution

It presents the 4N process, integrating deep learning with geostatistics, and develops new feature construction methods based on neighboring information.

## Key findings

- Outperforms existing geostatistical methods on simulated non-Gaussian data
- Validates the 4N process as a stochastic process
- Effective on large forestry datasets

## Abstract

Kriging is the predominant method used for spatial prediction, but relies on the assumption that predictions are linear combinations of the observations. Kriging often also relies on additional assumptions such as normality and stationarity. We propose a more flexible spatial prediction method based on the Nearest-Neighbor Neural Network (4N) process that embeds deep learning into a geostatistical model. We show that the 4N process is a valid stochastic process and propose a series of new ways to construct features to be used as inputs to the deep learning model based on neighboring information. Our model framework outperforms some existing state-of-art geostatistical modelling methods for simulated non-Gaussian data and is applied to a massive forestry dataset.