Crop Production Prediction Models in Indian Agriculture: Possibilities and Challenges

Abstract

Agriculture sector is an important contributor to Indian economy. Wheat yield has decreased from 3.53 kg ha (FY 2019) to 3.42 kg -1 ha (FY 2020). Crop production is affected by a lot of factors including climatological factors (temperature, precipitation), soil type and seed -1 quality. Crop production prediction is an important aspect for farmers, agro based industries and policy makers. Many technological tools based on data mining models are being developed to draw correlation agricultural datasets and crop yield. The prediction methodology involves learning the pattern of crop yield during a set of conditions based on the previous years' data and thus predicting the yield in the current set of environmental parameters. The present paper focuses on commonly used models for crop yield prediction in Indian agriculture including artificial neural network, bayesian belief network, support vector machine, decision tree regression analysis, random forest, least absolute shrinkage regression operator and elastic net. A comparison of their efficiency has been drawn based on the previous studies.