Nonparametric prediction with spatial data

TL;DR

This paper introduces a nonparametric spatial data prediction algorithm based on spectral density factorization, demonstrating its theoretical advantages and practical effectiveness through simulations and real estate data application.

Contribution

It presents a novel nonparametric prediction method for spatial data using spectral density factorization, with proven asymptotic properties and competitive finite sample performance.

Findings

The predictor has desirable asymptotic properties.

The method performs well in Monte Carlo simulations.

Successfully applied to house price prediction in Los Angeles.

Abstract

We describe a (nonparametric) prediction algorithm for spatial data, based on a canonical factorization of the spectral density function. We provide theoretical results showing that the predictor has desirable asymptotic properties. Finite sample performance is assessed in a Monte Carlo study that also compares our algorithm to a rival nonparametric method based on the infinite AR representation of the dynamics of the data. Finally, we apply our methodology to predict house prices in Los Angeles.