Machine Learning and Internet of Things Based Real Time NPK Fertilizer Prediction for Cassava Crop in Rwanda

This research paper focus on “Employing Machine Learning and Internet of Things Based Real Time NPK Fertilizer Prediction for Cassava crop in Rwanda”. Throughout this research study, a machine learning (ML) algorithm was used to build a model which can help to predict the cassava fertilizer components. Different sensors such as temperature sensor, soil moisture sensor, soil nutrients sensors, and PH sensor have been used for monitoring the soil parameters and then the data corresponding to these parameters applied to machine learning algorithms Linear Regression, Random Forest, Gradient Boosting, Random Forest, K-Nearest Neighbors and Decision Tree to be tested for optimizing the prediction accuracy and these algorithms have been selected because they are mostly used in classification and regression machine learning problems. During the model performance evaluation, 93.1% and 90% of training and testing prediction accuracies respectively were achieved by using random forest for the soil samples taken in Ruhango District located in Southern Provence of Rwanda. However, 96.5% and 94.4% of training and testing prediction accuracies were generated by using the decision tree predictive model.