Comparison of C4.5 Algorithm and K-Nearest Neighbors on the Classification of Multiple Intelligence Test Results for Recommended Student Lectures

Abstract

This research uses K-Nearest Neighbors and C4.5 classification algorithm aimed to classify student lecture field based on multiple intelligence test result. There are 2 categories of fields used in the classification, namely Science-Technology (SainsTech) and Social-Humanities (Soshum.) The data used obtained from multiple intelligence test conducted to students of Sebelas Maret University Surakarta. The collected data sets will be transformed and normalized for a classification process. The selection feature is performed to remove any inappropriate attributes.

The test is done by using confusion matrix, by doing 12 experimental scenarios with different datasets and classification techniques. From the experimental results, it is shown that the K-Nearest Neighbors algorithm is better than the C4.5 dataset of normalization result has the highest accuracy of 56.84% to 54.84%.