Evaluation of Crop Yield Prediction Using Arsenal and Ensemble Machine Learning Algorithms

Abstract

Agriculture remains the prime source of living which is the keystone of our country. Present challenges like water scarcity, unpredictable cost, and weather ambiguity require farmers to equip themselves to smart farming. In precise, the crop yield is low due to ambiguity in climatic changes, poor facilities in irrigation, decreased soil fertility, and traditional farming techniques. Farmers are cultivating the same crops frequently without testing a new variety of crops and they are using fertilizer without the knowledge of what quantity needs to be used which leads to uncertainty. Machine learning is a successful technique to answer these uncertainties. This article mainly aims to predict crop and its yield depending on historical data available alike weather, soil, rainfall and crop yield parameters alike soil PH value, temperature, and climate of the particular area using various arsenal and ensemble algorithms. A comparative evaluation of prediction based on these algorithms are presented. The proposed GUI application predicts the type of crop to be cultivated that can give high yield based on the parameters given by the user. This application can be widely used by farmers to grow variety of crops based on the constraints, and thus increase the profit of yield and can invariably decrease the soil pollution.