Ensemble of Classifiers Based Soil Evaluation for Crop Cultivation

This paper describes an intelligent method of soil evaluation for crops cultivation by using multi-class labels. The method assists farmers to increase their crop production and comprises a total of four steps: (i) soil data collection, (ii) prepossessing, (iii) soil feature extraction, and (iv) implementation of classifiers by using different machine learning algorithms. The applied machine learning classifiers are naive bayes, k-nearest neighbor, random forest, decision tree, support vector machine, and artificial neural network to classify soil features. Later we also build an ensemble of the previous classifiers. From the experiments, it is observed that the method based on ensemble approach (decision tree + random forest + multi-layer perceptron) provides the best accuracy of 98.03%, whereas the decision tree provides second highest accuracy of 87.09% over the other methods investigated.