CLASSIFICATION OF BLOOD DONOR DATA USING C4.5 AND K-NEAREST NEIGHBOR METHODS (CASE STUDY: UTD PMI BALI PROVINCE)

Abstract

Classification of blood donor data at UTD PMI Bali Province by applying the C4.5 and K-Nearest Neighbor algorithms. The number of blood donor data donors is 34,948, of which 90% of the data, namely 31,454 is used as training data. Meanwhile, 10% of the data, which is 3,494 data, is used as the implementation of data testing using the Orange application. C4.5 obtained an accuracy score of 92.9%, F1 of 92.2%, Precision of 93.1%, Recall of 92.9%, specificity of 68.2%. While K-nearest neighbor obtained an accuracy score of 91%, F1 of 90.1%, Precision of 90.8%, Recall of 91%, specificity of 63%. With the AUC (Area Under Curve) value for the C4.5 algorithm is 0.875 and the K-nearest neighbor is 0.813 Good Classification. The results of the evaluation using the confusion matrix C4.5 obtained an accuracy score of 92.6%, F1 of 95.7%, Precision of 99.4%, Recall of 92.4%, specificity of 96%. While k-nearest neighbor obtained an accuracy score of 90.9%, F1 of 94.6%, Precision of 98.4%, Recall of 91.2%, specificity of 88.4%. Based on the evaluation of the confusion matrix and the ROC Analysis Graph, the C4.5 algorithm obtained higher results than the K-Nearest Neighbor algorithm. Based on the data on the characteristics of blood donors at UTD PMI Bali Province, it shows that the gender is male, Badung area, Age 20 to < 30, the occupation of private employees dominates in blood donation.