Performance Evaluation of Machine Learning Regression Algorithms for Soil Nitrogen Estimation

Soil nitrogen is vital for healthy plant growth, and its proper balance is crucial for soil fertility. Machine learning (ML) has gained popularity in agriculture due to its ability to analyse large datasets quickly and accurately, offering advantages over traditional time-consuming and costly laboratory approaches. This study compares five ML regression algorithms, namely, multiple linear regression (MLR), decision tree regression (DTR), random forest regression (RFR), k-nearest neighbour regression (KNN R), and support vector regression (SVR) to predict soil nitrogen (N) levels using electrical conductivity (EC), organic carbon content (OC), and pH values of soil. The trained models are assessed using performance metrics such as root mean squared error (RMSE), r-squared (R2), and mean absolute percentage error (MAPE). Among these models, SVR demonstrates the most promising result, with an R2 value of 0.928, MAPE value of 15.90, and RMSE value of 1.053.