PERBANDINGAN GULUD REGRESSION DAN PRINCIPAL COMPONENT REGRESSION (PCR) TERHADAP PEMODELAN INDEKS PEMBANGUNAN MANUSIA PROVINSI JAWA TIMUR

Raihandika Ari Indhova (Departemen Statistika, Fakultas Sains dan Matematika, Undip)
Suparti Suparti (Departemen Statistika, Fakultas Sains dan Matematika, Undip)
Arief Rachman Hakim (Departemen Statistika, Fakultas Sains dan Matematika, Undip)

Jurnal Gaussian
Vol 12, No 2 (2023): Jurnal Gaussian

Human resources is valuable asset in a country. Human Development Index (HDI) becomes important indicator of quality of human resources in an area. HDI value is affected by a variety of factors that are strongly related to each other so they cause multicollinearity. This observation aims to deal with multicollinearity optimally by comparing Gulud Regression to Component Regression in modeling factors that affect East Java HDI in 2020. Data that are used in this observation are East Java HDI in 2020 (Y), Life Expectancy (X1), Infant Mortality Rate (X2), Mean Years of Schooling (X3), Expected Years of Schooling (X4), Open-Unemployment Rate (X5), Average Household Expenditure per Capita (X6), and Labor Force Participation Rate (X7). Based on MSE value, the Gulud Regression method is better than Principal Component Regression (PCR) method in dealing with multicollinearity problem. Based on adjusted  score that is 0,954, feasibility test of the best model of Gulud Regression method is a strong model.