Crop Fertilizer Prediction using Regression analysis and Machine Learning algorithms
K Monika, Balakrishnan Ramprakash, Sankayya Muthuramalingam, K Mirdula
Abstract
India is one among the well-known countries whose socio-economic status is dates by agriculture. The rise of sedentary human civilization is the key role achieved in the development of agricultural practice. The problems in agriculture revolves not only about the productivity of the crop but also the amount of fertilizer, herbicide, insecticide used and also include the amount of water that are required to get a good productivity to the farmers. Machine Learning is used in agriculture to find solutions to the problems with high accuracy for the benefit of the farmers by recommending fertilizers, pesticides, etc., according to their needs. There are several machine learning algorithms that are widely used in our day-today life in various domains. This paper gives a detailed description of predicting the suitable variety of fertilizer based on the machine learning technique like multiple linear regression and lasso regression as it takes the relationship between the dependent and independent variables that affects the resultant prediction. It gives a brief overview for recommending suitable fertilizer for the conditions given by the farmers like temperature, humidity, soil type, crop type, quantity of fertilizers like nitrogen, potassium, phosphorus that are present in the soil, etc., This paper implies that the lasso regression is the most suitable regression algorithm for predicting the suitable variety of fertilizer for the climatic and field conditions given by the farmers with an accuracy of 89%. The developed model helps the farmers to find the most suitable fertilizer with respect to the soil and climatic conditions of the field to have higher yield, which in turn increases the profit for the farmers. The application of the suitable variety of fertilizer to the soil also helps to increases the nutrient contents that are present in the soil.