Consequences of spatial structure in soil-geomorphic data on the results of machine learning models

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00027073%3A_____%2F23%3AN0000016" target="_blank" >RIV/00027073:_____/23:N0000016 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/62156489:43410/23:43923822

Výsledek na webu

<a href="https://www.tandfonline.com/doi/full/10.1080/10106049.2023.2245381" target="_blank" >https://www.tandfonline.com/doi/full/10.1080/10106049.2023.2245381</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1080/10106049.2023.2245381" target="_blank" >10.1080/10106049.2023.2245381</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Consequences of spatial structure in soil-geomorphic data on the results of machine learning models

Popis výsledku v původním jazyce

In this paper, we examined the degree to which inherent spatial structure in soil properties influences the outcomes of machine learning (ML) approaches to predicting soil spatial variability. We compared the performances of four ML algorithms (support vector machine, artificial neural network, random forest, and random forest for spatial data) against two non-ML algorithms (ordinary least squares regression and spatial filtering regression). None of the ML algorithms produced residuals that had lower mean values or were less autocorrelated over space compared with the non-ML approaches. We recommend the use of random forest when a soil variable of interest is weakly autocorrelated (Moran's I < 0.1) and spatial filtering regression when it is relatively strongly autocorrelated (Moran's I > 0.4). Overall, this work opens the door to a more consistent selection of model algorithms through the establishment of threshold criteria for spatial autocorrelation of input variables.

Název v anglickém jazyce

Consequences of spatial structure in soil-geomorphic data on the results of machine learning models

Popis výsledku anglicky

In this paper, we examined the degree to which inherent spatial structure in soil properties influences the outcomes of machine learning (ML) approaches to predicting soil spatial variability. We compared the performances of four ML algorithms (support vector machine, artificial neural network, random forest, and random forest for spatial data) against two non-ML algorithms (ordinary least squares regression and spatial filtering regression). None of the ML algorithms produced residuals that had lower mean values or were less autocorrelated over space compared with the non-ML approaches. We recommend the use of random forest when a soil variable of interest is weakly autocorrelated (Moran's I < 0.1) and spatial filtering regression when it is relatively strongly autocorrelated (Moran's I > 0.4). Overall, this work opens the door to a more consistent selection of model algorithms through the establishment of threshold criteria for spatial autocorrelation of input variables.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10511 - Environmental sciences (social aspects to be 5.7)

Projekt

<a href="/cs/project/SS02030018" target="_blank" >SS02030018: Centrum pro krajinu a biodiverzitu</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Geocarto International

ISSN

1010-6049

e-ISSN

1752-0762

Svazek periodika

38

Číslo periodika v rámci svazku

1

Stát vydavatele periodika

GB - Spojené království Velké Británie a Severního Irska

Počet stran výsledku

15

Strana od-do

2245381

Kód UT WoS článku

001048405500001

EID výsledku v databázi Scopus

2-s2.0-85168159365