2-Stage Hybrid Ensemble-Based Heterogeneous Committee Machine for Improving Soil Fertility Status Prediction Performance

Adequate and accurate information about the soil is one of the key basic fundamental factors for optimal decision-making and sustainably improving crop productivity. Contemporarily, advanced machine learning modelling techniques have been useful in delivering actionable information in various fields. In agriculture, soils in particular, the same has been the case in various tasks including soil nutrients analysis and fertility status predictions, as well as determination of appropriate crops for plantation, among others. There exists much exploration in regression models for soil nutrients, compounds, and their key chemical characteristics. However, varying predictive performances have been demonstrated by ML models for predicting soil fertility status. In addition, a range of varying fertility status levels have been synthesized and used for modelling classifiers targets for predicting soil fertility. This paper presents a high-performance throughput 2S-HHEC combining a neural network, decision tree, random forest, support vector machine, and gradient boosting classifiers to predict soil fertility status with accuracy and Kappa score of 98.9% and 93.9% on test data.