Machine Learning Techniques For Crop Yield Prediction

Abstract

For many developing nations, agriculture is the fundamental source of economic engine. Modern agricultural advancements guarantee great crop yield by using better farming methods. Farmers find it difficult to meet the changing needs of the earth. They toil day and night to ensure a plentiful harvest at session ending. The availability of water, soil fertility, the prevention of rodent damage to crops, the timely use of pesticides and other helpful chemicals, and the presence of nature all contribute to a healthy harvest. A farmer can regulate the quantity and frequency of pesticide applications, even though many of these other elements are difficult to manage. Pesticides are crucial to the development of agricultural products. They have been used by farmers to control weeds and insects, and it is reported that they have contributed significantly to increased agricultural output. Without a significant increase in food production, the rise in global population over the twenty-first century would not have been conceivable. Pesticide use determines the production
of almost one-third of agricultural products. The precise prediction of healthy crop output using machine learning algorithms is thus a crucial problem in agricultural growth. To establish an accurate and effective model for crop classification, including crop yield estimate based on weather, crop disease, classification of crops based on the growth phase, etc., several investigations are advised. In this paper, the goal is to find the best model that can assist farmers in determining the degree of crop damage based on the type of crop, soil, pesticide usage, etc. Five popular algorithms such as K-Nearest Neighbor (KNN), Random Forest (RF) Classifier, Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM) and Multi-layer Perceptron (MLP) are compared on the crop dataset using the metrics mean accuracy, standard deviation and boxplot. Experimental results reveal that XGBoost gives the highest accuracy among the five, while MLP and LightGBM show delicate difference with XGBoost.