ANALISIS KOMPARASI PEMODELAN ALGORITMA DECISION TREE MENGGUNAKAN METODE PARTICLE SWARM OPTIMIZATION DAN METODE ADABOOST UNTUK PREDIKSI AWAL PENYAKIT JANTUNG

Abstract

Cholesterol, fat accumulation, blood clots, and other medical elements can cause
blood flow to the heart to be disrupted so that this medical condition becomes an
emergency that destroys and destroys the heart muscle in humans resulting in
complications and even death. This research will improve the Classification Accuracy
/ Ensemble Methods, Techniques by modifying the Decision Tree classification
algorithm modeling, which is added by the Particle Swarm Optimization method and
Adaboost method which then will be comparative analysis to the modification for
heart disease prediction. The dataset used in this study from the extraction of public
dataset taken from the archive of the University of California Irvine (UCI) of 270
records. Evaluation result done by Rapid Miner 7 tool to determine the confusion
matrix and ROC curve value, it is known that Decision Tree has 79.26% accuracy and
for AUC 0.889. After modifying the decision tree modeling algorithm added with the
particle swarm optimization method resulted in an accuracy of 82.59% and for AUC
0.916. The decision tree decision algorithm added with the adaboost method yields an
accuracy of 79.26% and an AUC value of 0.955.