Analysis and Evaluation of Crop Yield Pattern Based on Various Forecasting Methodologies

Agriculture is the backbone to the Indian financial state. It makes up about 17% of overall GDP (Gross domestic product) and offers employment to nearly 58% among the populace. But there are many challenges in achieving the precision in the agricultural activity related with estimation and production of crops, which includes crop and weed detection, uncertain water and atmospheric conditions, biomass evaluation and yield prediction. In the proposed study, the emphasis has been laid down on predicting the crop yield pattern based on some prominent features such as: ratio of nitrogen, phosphorous and potassium in the soil, humidity, rainfall, temperature, and pH of soil, which directly affects the pattern of crop yielding. In retrospect, there are numerous machine learning methodologies that have been proposed to evaluate their performance with respect to estimated and targeted production of several crops. In this proposed work, a comprehensive comparative analysis of the most significant machine learning classifiers has been done for recommending the crop name (which crop to grow). The prominent classifiers analyzed in this work are Logistic Regression, Naïve Bayes, Decision Tree, Random Forest, SVM and KNN.