Development Of Efficient Wheat Crop Yield Prediction Technique Based On Different Environmental Parameters

Abstract

Agriculture plays a vital role in the economy of a nation. The changing environmental newlineconditions and the alarming growth in the population is constantly posing a challenge for the newlinefarmers to meet the increasing demand of food crops all over the world. Wheat is one of the newlinemost consumed staple food crop of the world. The unprecedented increase in environmental newlinetemperature and reduction in amount of rainfall is quite detrimental to the growth and newlinecultivation of this heat sensitive crop. The researchers all over the world are trying to find ways newlineof increasing the production of wheat crop with minimal destruction to the natural resources newlinecalled Climate Sustainable Agriculture Practices. An accurate and timely prediction of crop newlineyield before harvest can be of great help for the researchers and farmers to prior assess the risk newlineand take due measures to ensure stable yield of crop. Broadly, crop yield prediction models newlinecan be categorized as crop growth models and data driven models. Crop growth models are newlineefficient way of crop yield prediction but are time consuming, quite expensive and less accurate newlinedue to varying environmental conditions. Hence, the timely action to improve the crop yield is newlinenot in the scope of the farmer. Data driven models are less expensive empirical models and the newlineemergence of machine learning algorithms further added efficacy of these models. Inspite of newlineextensive advancements in the field of machine learning and deep learning, the techniques newlinehave not been fully utilized for an accurate crop yield prediction. The focus of the present newlineresearch is to propose an efficient crop yield prediction technique for an accurate and timely newlineyield prediction of wheat crop for one of the Punjab regions of India. A hybrid deep learning newlinemodel, RNN with LSTM, is proposed for an accurate and timely prediction of wheat crop yield. newlineThe study has also optimized two important hyperparameters of the RNN-LSTM model; newlinewindow size and number of neurons in hidden layer, to increase the efficacy of the model using newlinegenetic algorit