Novel hybrid methods applied for spatial prediction of mercury and variable selection of trace elements in coastal areas of USA
2020
Sakizadeh, Mohammad
This study was concerned with spatial analysis of mercury (Hg) in sediment samples of the USA coastal areas using more accurate and stable hybrid approaches compared to the conventional methods. An ensemble of simulated annealing along with least angle regression (SA-LAR) was applied for selection of predictors in spatial analysis. The latest algorithm was efficient with resultant RMSE and R² of 0.066 and 0.705 compared to 0.099 and 0.571 for the traditional method of recursive feature elimination (RFE) approach. Using Cu, Pb and As as selected variables, it was tried to improve the spatial forecasting of Hg with either a hybrid of generalized boosted regression and ordinary kriging (GBROK) or inverse distance weighting (GBRIDW). According to the results, the variance explained by cross validation (VECV) was improved from 7.52% and 9.76% for IDW and OK to 40.41% and 41.94% for the GBRIDW and GBROK methods, respectively.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Cette notice bibliographique a été fournie par National Agricultural Library
Découvrez la collection de ce fournisseur de données dans AGRIS
