Soil NPK Prediction Using Multiple Linear Regression

Abstract

Soil nutrients are the important parameter which contributes a major role in healthy plant growth. In soil, the presence of three macro nutrients namely Nitrogen (N), Phosphorus (P), and Potassium (K) are essential for proper crop growth. Without having adequate knowledge about nutrient levels present in soil, farmers apply large quantity of fertilizers in their field. This leads to depletion or enhancement of nutrient content in the soil and it degrades the soil fertility. Laboratory soil test is time consuming and it involves addition of many chemical reagents. Hence in this paper, the soil macronutrients are predicted using Multiple Linear Regression (MLR) technique. It is a statistical method where several independent or explanatory variables are used to predict the dependent or response variable. MLR technique is formulated to determine a mathematical relationship among several parameters. It shows the relationship between dependent and independent variables. In this technique, soil parameters like nitrogen, phosphorus, potassium, pH and electrical conductivity are used to sketch the relationship among these parameters and predict the values of NPK. The predicted NPK data shows an accuracy of approximately 80% when compared with the actual dataset. These results improve the decision-making capabilities of farmers in applying right quantity of fertilizers and increase crop production.