pone.0251952.g011.tif (803.85 kB)
Classification result.
figure
posted on 2021-12-16, 18:45 authored by Santosh Hiremath, Samantha Wittke, Taru Palosuo, Jere Kaivosoja, Fulu Tao, Maximilian Proll, Eetu Puttonen, Pirjo Peltonen-Sainio, Pekka Marttinen, Hiroshi MamitsukaObserved (left) vs predicted (right) crop loss of 1000 randomly chosen field parcels in 2008 simplified as points. Red point indicates croploss of the fieldparcel, blue point indicates no loss in 2008.
History
Usage metrics
Keywords
using satellite imagesusing machine learningtrain machine learningnumerous potential applicationslandsat 7 imagesfinnish food authorityfield parcel scaleallowing government agencies688 &# 1775 million haidentifying new cropextremely noisy datawithout crop lossdefine crop losscrop loss identificationloss fields basedmissing value indicatorscrop loss classificationreliable reference datafour ml modelscrop lossreference fieldsreference dataverify croploss claimspossible dataclassification settingtrained modelsroc curverandom forestmean imputationmany waysmany factorsinsurance companieshard problemgrowing seasonfinland coveringaverage aucapproximately 116 years
