Predicting the Behavior of Blood Donors in National Blood Bank of Ethiopia Using Data Mining Techniques

A modern technology used for extracting knowledge from a huge amount of data using different models and tasks such as prediction and description is called data mining. The data mining approach has a great contribution on solving a different problem for data miners. This paper focuses on the application of data mining in health centers using different models. The model development process helps to identify or predict the behavior of blood donors whether they are eligible or ineligible to donate blood by their right status way and protects any blood bank health center from the collection of unsafe blood. Classification techniques are used for the analysis of Blood bank datasets in this study. For continuous blood donors, it will help to enable to donate voluntary individuals and organizations systematically. J48 decision tree, neural network as well as naïve Bays algorithms have been implemented in Weka to analyze the dataset of blood donors. The study is used to classify the blood donor's eligibility or ineligibility status based on their genders, deferral time, weight, age, body priced, tattoos, HIV AIDS, blood pressure, donation frequency, hepatitis, illegal drug use attributes. From the 11 attributes, gender does not affect the result. We have used 1502 datasets for the train set and 100 datasets for testing the model using cross-fold validation. Cross-fold data, partition was used in this study. The efficiency and effectiveness of the algorisms are measured automatically by the system. The obtained result showed that the J48 classifier outperforms the best result as well as both neural network and navies, Bayes, in terms of matrix evolution, with its 97.5% overall model accuracy has offered interesting rules.