Traits preferred by birds to attack on standing wheat crop under temperate condition and its prediction through random forest model

Random forest model-based prediction on bird pest attack on wheat

Abstract

Wheat is important cool season cereal crop of India, Production and quality get compromised due to biotic and abiotic stress. Wheat may be attacked by several disease and insect pests including birds. Birds attack on standing crop are not very often and extent and pattern of damage varies considerably, however, its management practices is generally not taken in to account. The bird preferred to attack wheat crop at reproductive stage especially during milking to dough stage. Data generated during the course of two-year field study data confirm that under Kashmir valley condition at Research Farm Srinagar (Jammu & Kashmir) has been subjected to Random Forest model to find out trait specific preference for attack if any. Random forest (RF) model identified five major features out of 13 selected features of wheat germplasm with seventy percent accuracy for parrot birds’ prefer to attack on wheat germplasm. RF Model suggest that tallerheight, having long spikes and a greater number of ears, along with more biomass and grain yields could be possible traits bird pest attack during reproductive stage. It was inferenced that the average performance of the random forest model demonstrated with model accuracy (71%), sensitivity (0.77), kappa 0.41 and ROC (0.74) and showed good fit for classification of parrot attack on wheat germplasm.