A Predictive Linear Model for Outbreak of Cocoa Blackpod Disease in South West Nigeria

Abstract

The aim of this study is to develop a predictive linear model that notifies farmers about the outbreak of cocoa blackpod disease in South-West Nigeria. Relevant dataset of precipitation and temperature which covers five cocoa producing states (Ondo, Ekiti, Osun, Ibadan and Ogun) was collected from Nigeria Meteorological Agency (NIMET) and it spans from 1988-2018 (30 years’ dataset). The predictive model was formulated with SARIMA, which was used in training the dataset and making predictions. The model was serialized with a mobile application developed using Application Programme Interface (API) so that farmers can receive monthly notification of blackpod disease. The proposed model was simulated using the python programming language. The predictive linear model was evaluated using the performance parameters: accuracy, precision, recall and Mean Square Error. Prototype implementation of the model was done with python programming language. The following evaluation results were derived from SARIMA: Accuracy: 0.8333; Precision: 0.6316; Recall: 0.6316; and Mean Square Error: 0.4082. The key contribution of this research work is the provision of a model that provides early and reliable information on the outbreak of blackpod disease. Another upgrade that was done in this research work included development of a mobile application for automatic farmer notification of possible blackpod infection through the mobile or Internet network.