Application of Data Mining Technology to Support Adequate Chemical Fertilizer Prediction for TEf And Wheat Production In Some Selected Parts of Ethiopia

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title


Addis Ababa University


Though agriculture is the mainstay of Ethiopian economy, it is suffering from many disasters. Among these disasters, nutrient depletion of the soil is the major one. Applying organic and/or inorganic (chemical) fertilizer in the soil can curb nutrient depletion. Scarcity of organic fertilizer, here in Ethiopia, brings about the need to use chemical fertilizer. But, still, there is a problem of using sufficient amount of chemical fertilizer based on initial fertility status of the soil and nutrient requirement of crops to bring high yield. Hence, this and others arouse interest of developing a guideline for fertilizer recommendation. This thesis developed a decision support system that can help agricultural researcher in the process of building a guideline for fertilizer recommendation. In doing so, the research aimed to assess the potential applicability of data mining technology specifically decision tree teclmique to help in fertilizer-grain yield data analysis in decision-making process. In this research, in the process of building a model, different steps were undertaken. Among the steps, data collection, data preprocessing and model building and validation were the major ones. Different tasks performed in each step are mentioned as follows. The data were collected from National Soil Testing Center. Under preprocessing, data cleaning, discrimination and attribute selection were done. The final step was model building and validation and it was performed using the selected tools and techniques. The data mining tool used in this research was Weka. In this software the decision tree 148 algorithm was selected since it is capable to analyze numeric data. After successive experiments were done using this software, a model that can classify crop yield as high, medium and low with better accuracy to the extent of 85% and sound rule was selected. Experimental results show that deci sion tree is a very helpful tool to depict the contribution of soil-pH, initial available so il phosphorus, organic-matter, total nitrogen and treatment to bring high tef and wheat yield. The reported findings are optimistic, making the proposed model a useful tool in the decision making process. Eventually, the whole research process can be a good input for Further in-depth research.



Information Science