OPTIMASI PSO PADA NAÏVE BAYES UNTUK PREDIKSI CALON PENDONOR DARAH TETAP (STUDI KASUS PMI KABUPATEN KUDUS)

Abstract

Blood donation is one of the crucial activities and one of the most important aspects to support the life safety of hospital patients who are in need of blood transfusions. However, in practice, the demand for blood bags is always higher than the number of blood bags produced, resulting in the availability of blood bags which are always empty and increasing the risk of patients who are not helped. Departing from this problem, a data mining classification technique is needed to predict prospective blood donors. The initial dataset obtained was 911 records in the period January to March 2023 with 18 (eighteen) attributes. Then data preprocessing is carried out so that the dataset will use 7 (seven) attributes with details of 1 (one) ID attribute, namely name, 5 (five) regular attributes, namely age, sex, blood group, hemoglobin, weight and 1 (one) attribute class i.e. routine donor status. The research uses the Naïve Bayes method which is optimized using Particle Swarm Optimization with testing using a confusion matrix to produce accuracy and AUC values with rapidminer software analysis. an increase of 5.31% so that it rose to 80.41%. An AUC (Area Under Curve) value of 0.716 was obtained which was categorized as a fair classification. By generating knowledge in the research process, in the future it will be able to overcome the gap in supply of blood bags at PMI Kudus Regency.

Keywords: Prediction, Blood Donation, Naïve Bayes , Particle Swarm Optimization