PENGGUNAAN ALGORITMA C4.5 DAN LOGIKA FUZZY UNTUK KLASIFIKASI TALENTA KARYAWAN (Studi Kasus : Politeknik Negeri Bali)

Human resources analysis, especially at the competency based talent analysis, is the important thing to do by the institution for their efficiency in human performance employee. Many institutions try to do talent analysis to do the right staff recruitment that have good capabilities. That�s why Ba...

Full description

Bibliographic Details
Main Authors: , NI G. A. P. HARRY SAPTARINI, , Drs. Retantyo Wardoyo, M.Sc., Ph.D.
Format: Thesis
Published: [Yogyakarta] : Universitas Gadjah Mada 2012
Subjects:
ETD
Description
Summary:Human resources analysis, especially at the competency based talent analysis, is the important thing to do by the institution for their efficiency in human performance employee. Many institutions try to do talent analysis to do the right staff recruitment that have good capabilities. That�s why Bali State Polytechnic (PNB) also applies this analysis. To analyze the human resources competency can be done by analyzing the human talent. One of the conventional method that common to use is C4.5 algorithm. Conventional C4.5 method uses data crisp. In this study, it used a linguistic term as an input data because the talent test is expressed using the language (linguistic term) and it is a set of data in fuzzy form. Fuzzification process in preprocessing data was done to produce the input data in fuzzy form. The results shows the use of fuzzy logic in a stage of preprocessing and the C4.5 algorithm in forming decision tree (fuzzy C4.5) can be used as an alternative solution to process the input dataset numeric format, where the number of linguistic terms of an attribute directly and significantly affect the accuracy of the system. The more the number of linguistic terms, the lower the accuracy of the system. The result of this research is obtained in the use of fuzzy logic in that stage of pre-processing and algorithms C4.5 in forming the decree ( fuzzy C4.5 ) can be used as solution alternative to cultivate dataset inputs numerical, where's the number of linguistic term of an attribute influential directly and significantly to the accuracy of the system . The more linguistic term so the lower the accuracy of the system