Abstract: Insect pests and plant diseases are one of the major causes for decrease in the quality of agricultural productivity and are extending quickly across the region. This is a critical matter for the continent and especially for South Africa as it jeopardises a crucial resource for economic growth and ecosystem conservation. This project determines prevalence of diseases and investigates supervised learning algorithms for the classification of diseases and pests that affect trees. The proposed method uses geo-coding to first understand the location of the different diseases, applies text-analytics and feature engineering for re-structuring data, and then provides a classification of the diseases using three machine learning algorithms: linear discriminant analysis, decision trees and random forests. The experimental results demonstrate that random forests are the most efficient with an accuracy of 66%.