The Effect of Particle Swarm Optimization Feature Selection in Predicting Blood Donor Eligibility Classification at UTD Bekasi City Using Decision Tree C4.5

Fulfilling the need for blood is very important to improve the quality of health services in a country, one of the Puskesmas in Indonesia is dedicated to helping people to donate blood called a Blood Transfusion Unit. In carrying out blood donation, the obstacle that UTD officers often face is when the demand for blood increases or the number of donors is large. Of course, it takes a long time to check the eligibility of donors, sometimes it makes donors bored to the point of canceling their determination to donate for reasons of time. This is due to the limited number of officers involved and still using manual methods in determining donor eligibility. Thus, an efficiency model is needed to check the eligibility of blood donors to maximize service. In this study utilizing data mining to determine the eligibility status of donors at UTD PMI Bekasi City, the algorithm used is Decision Tree C4.5 with PSO as an attribute selector. The research results obtained using Decision Tree C4.5 only in classification obtained an accuracy of 96.46%. Then, testing was carried out using the Decision Tree C4.5 algorithm with PSO optimization increasing the accuracy value of 97.46%. In conclusion, the use of the PSO algorithm can improve the training performance of the Decision Tree C4.5 classification model. Based on the resulting decision tree, the hemoglobin attribute is the root of the decision tree, so hemoglobin is the most influential factor in determining the classification of blood donor status.