PJB-2019-771
IDENTIFICATION OF YIELD PREDICTORS IN WHEAT (Triticum aestivum L.) UNDER SALT STRESS USING RANDOM FOREST, MULTIPLE REGRESSION AND STEPWISE REGRESSION
Md. Hasanuzzaman
Abstract
Salinity is a major constraint for wheat production in the coastal area in Bangladesh. This research was conducted with ten wheat genotypes in earthen pots with control and 10 ds-1 salinity following completely randomized design with three replications from 2013 – 2014 to 2014 - 2015. Data were recorded on days to heading, number of effective tillers per plant, days to maturity, plant height, spike length, spikelets per spike, grains per spike, root length, root volume, root weight, root dry weight and yield per plant. Salinity represses the development of roots causing yield loss. Random forest, multiple linear regression and stepwise regression, all three methods have identified dry root weight and number of effective tillers per plant play important role in yield per plant under salt stress. Selection through these traits may be effective in saline environment. Random forest performs better than multiple linear regression and stepwise regression models showing lowest MSE and R2.
To Cite this article:
Download
