Agriculture is the backbone of the Ethiopian economy and it contributes the highest GDP of the country. Among this, crop production takes the highest level of income for most smallholder farmers in all regions of Ethiopia. The objective of this research is to build a model that can predict crops productivity and implement a decision support system. In order to conduct this research, a hybrid Knowledge Discovery Process model was adopted. For the purpose of this research, the datasets were taken from Central Statistical Agency of Ethiopia database, and the researcher used a total of 25,000 instances for training and building a model. Hence, for building a model and implementing decision support system for predicting crop productivity, WEKA data mining tool and java NetBeansIDE was used respectively. To achieve the objective of these research different experiments were conducted using J48, HoeffdingTree decision tree and PART rule based classifiers. In addition, the predictive performances of the classifiers are evaluated and compared using accuracy rate, confusion matrix and ROC curve. Based on this, out of the three classifiers PART rule based classifier performs best accuracy and ROC rate which is 95.44 % and 0.992 respectively. As a result PART rule based classifier were selected for implementing the model to predict crop productivity. In this thesis, the experimental result shows that, the main determinant factors for crop productivity are main season (season type), use of extension program, fertilizer used and fertilizer type. Therefore, the outcome of this research is essential to make data mining based decisions for policy makers and for experts in the area of crop agriculture to give an attention on the factors affecting crop productivity and to take corrective measures.

Data mining predictive model decision support system crop production Ethiopia