Comptes rendus de l’Acade'mie bulgare des Sciences, Vol 71, No9, pp.1279-1287

Adaptive Lasso Analysis for Grain Yield and Yield Components in Two-rowed Barley under Rainfed Conditions

Suna Akkol, Diğdem Arpalı, Mehmet Yağmur

ABSTRACT

The goal of this study was to determine the yield components related to grain yield in order to improve barley yield under rainfed conditions of Turkey (Van). Stepwise and Adaptive Lasso methods were performed for selection of most significant yield components. As cohesion criteria to compare Stepwise and Adaptive Lasso methods, the adjusted coefficient of determination and Akaike Information Criterion were used. Results revealed that when there were dependencies between independent variables stepwise and Adaptive Lasso achieved the same results. It has been determined that spike number per m2 and grain weight per spike can be used as the most effective selection criteria for barley breeding studies due to their significant effects on grain yield.

Key words: lasso regression, variable selection, shrinkage methods, tworowed barley, grain yield

DOI: 10.7546/CRABS.2018.09.17

Topic: AGRICULTURAL SCIENCES