Integrated Predictive Model and Knowledge Based System for Wheat Disease Detection: Case of Arsi Zone

Abstract

Ethiopian economy is Agriculture dependent as 80% of the country's population reliant on agriculture. Wheat covers the country's 26% of annual cereal crop production and supplying nutritional benefits. The recent outbreak of cereal crop diseases resulted in the loss of productivity throughout the country. There is a need for efforts of all stakeholders in agriculture. Data mining and KBS come across with varieties approaches and application in solving agriculture problems. The main objective of the study is to investigate the construction of a predictive model to extract hidden knowledge and integrate with the Knowledge Base System for detecting Wheat diseases. This could help in the effort of enhancing decision making in the agriculture sector. For this purpose Mixed research design, hybrid knowledge acquisition techniques, prototyping system development method, rule based knowledge representation, WEKA with JDK 8, and Java libraries like JavaFX, JESS and Eclipse have been selected and implemented to achieve the objective of the study. To identify the best prediction model for detection of wheat diseases four experiments were undertaken with four respective classification algorithms. Objective evaluation standards were used to evaluate the prediction accuracy and performance of implemented classifiers. As a result, JRip classification algorithm selected and integrated in the development of KBS since, it registered better performance result of 99.61 % accuracy with Objective evaluation method. In addition to this, system performance and users’ acceptance testing have been undertaken to evaluate the prototype KBS. Test cases and questionnaire were prepared to evaluate the performance and users’ acceptance of the developed system prototype. As a result, the system can achieve in the absence of domain experts 80 % system performance evaluation result; indicating the KBS has the necessary knowledge for detection and treatment of wheat diseases which in turn implies that the result of study was applicable and operative. Besides to this, the proposed system achieves 86.3% of the users’ acceptance which indicates it could be operational, if implemented. Since the study achieved better results, it can be applicable for stated objective in araea.