Predicting methane yield by linear regression models: A validation study for grassland biomass

Vasilis Dandikas, Hauke Heuwinkel, Fabian Lichti, Jörg E. Drewes, Konrad Koch

Publikation: Beitrag in FachzeitschriftArtikelBegutachtung

17 Zitate (Scopus)

Abstract

The objectives of this study were to assess and validate previously published prediction models with an independent dataset and to expose the power and limitations of linear regression models for predicting biomethane potential. Two datasets were used for the validation, one with all individual samples and one with the average values of each cultivar. The results revealed similar performances of all four models for the individual samples. The methane yields of the cultivars were predicted more accurately than the methane yields of the individual samples. The grassland specific model predicted the variation in the dataset with an R2 of 0.84 and the slope of the regression line was equal to 1.0. Linear regression models are suitable to depict the variation in methane yield and for substrate ranking. However, the prediction error of the absolute values may be high since systematic external effects cannot be determined by a regression model.

OriginalspracheEnglisch
Seiten (von - bis)372-379
Seitenumfang8
FachzeitschriftBioresource Technology
Jahrgang265
DOIs
PublikationsstatusVeröffentlicht - Okt. 2018

Fingerprint

Untersuchen Sie die Forschungsthemen von „Predicting methane yield by linear regression models: A validation study for grassland biomass“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren