Implementasi Algoritma Genetika Untuk Menentukan Variabel Bebas Dalam Prediksi Nilai Skripsi Menggunakan Regresi Linier Berganda

Studi Kasus : Jurusan Manajemen, Universitas Maritim Raja Ali Haji

Authors

  • Silha Wildania Utami Universitas Maritim Raja Ali Haji
  • Martaleli Bettiza Universitas Maritim Raja Ali Haji
  • Muhamad Radzi Rathomi Universitas Maritim Raja Ali Haji

Keywords:

Multiple Linear Regression, Genetic Algorithms, Prediction, Thesis Value

Abstract

Thesis is a scientific work of students about problems that are in accordance with the competence of study programs as one of the requirements to obtain a Bachelor's degree (S-1). Thesis is considered important because it is the final evaluation of students in pursuing their studies at tertiary institutions. One indicator of students' success in completing a thesis is their understanding of the subjects that support their thesis. Therefore, we need a system that can predict the value of the thesis in order to become a benchmark and assist in the process of improving the quality of learning. From these problems, this study uses multiple linear regression to predict the value of thesis and genetic algorithms are used to determine the independent variables, which are the subjects that affect the value of the thesis. The reproduction process uses one-point-crossover and random mutation, for the selection process using the tournament selection model. From the trial, the best parameters are population size of 60, the best generation of 125 generations, the combination of cr: mr is 0.9: 0.9. Based on the results of four experiments of genetic algorithm methods - multiple linear regression gets an average MAE error value of 2.7620920388058.

Published

2020-05-11

Issue

Section

Teknik Informatika