Perbandingan Akurasi Metode Principal Component Analysis (PCA) dan Correlation-Based Feature Selection (CFS) Pada Klasifikasi Perpanjangan Kontrak Karyawan Menggunakan Metode Naïve Bayes

Abstract

PT. Oasis Waters International Palembang conducts regular staff performance reviews, the findings of which are utilized to make recommendations for employee contract extension. The Human Resource Department has assigned a numerical value to 25 qualities (HRD). The process of giving a label or class to a number of examples when the value of each characteristic is known as classification. The Naïve Bayes technique is a basic classification approach that makes use of probability estimates. Based on the observations, it was discovered that one of the 25 criteria was deemed the most relevant in determining the recommendation for an employee contract renewal. As a result, in this study, a comparison of the pre-processing Principal Component Analysis (PCA) approach and the Correlation-based Feature Selection (CFS) method on the categorization of employee contract extensions at PT Oasis Waters International Palembang will be performed. According to the data, the CFS approach has a positive influence on classification performance, while PCA does not. This is demonstrated by a 30% increase in accuracy when utilizing the CFS approach. Meanwhile, both strategies have a positive influence on the model's dependability. This is demonstrated by a reduction in Root Mean Square Error (RMSE) when using the CFS approach from 0.6325 to 0.1845, whereas using the PCA method results in 0.5123.Keywords : Naïve Bayes, Principal Component Analysis, Correlation-based Feature Selection, Confusion Matrix, Root Mean Square Error