PEMODELAN PENGELUARAN PER KAPITA DAN PERSENTASE PENDUDUK MISKIN DI JAWA TENGAH MENGGUNAKAN REGRESI BIRESPON SPLINE TRUNCATED

*Merinda Pangestikasari  -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Rita Rahmawati  -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Dwi Ispriyanti  -  Departemen Statistika, Fakultas Sains dan Matematika, Universitas Diponegoro, Indonesia
Received: 12 Feb 2020; Published: 13 Feb 2020.
View
Open Access Copyright 2020 Jurnal Gaussian
License URL: http://creativecommons.org/licenses/by-nc-sa/4.0

Citation Format:
Abstract

The Central Bureau of Statistics states that the average per capita spending (Y1) of Central Java Community in 2016 is around 27.808 rupiah per day. This value is still considered low, because it covers all the needs of an individual's life. The low expenditure per capita indicates the low level of welfare. Another indicator that can be used to measure community welfare is the percentage of poverty (Y2). Through this variable can be known how proportion of people who still difficult to meet their needs. Many factors are suspected to affect welfare, one of which is the average variable of school length (X). This study aims to get the best model and know the goodness of the model. Approach is done by nonparametric regression that is regres biresponse spline truncated. Nonparametric approach is done when data function does not show certain pattern. The best spline truncated biresponse model is highly dependent on determining the order and location of the optimal knot point that has a minimum Mean Square Error (MSE) value. In this study, the best model is obtained when order of Y1 is 2 and order of Y2 is 2 with five knots. The location of the knot point obtained is 7,05; 7,17; 7,32; 9,82 and 10,29 with MSE value of 662634,2. The goodness of the model is measured based on R-Square and MAPE, R-Square=43,21%, means the variance of response variables that can be explained by the predictor variable are 43,21% while the rest is influenced by other variables and MAPE=14,25%. Based on the value of MAPE can be said that the model had a good performance.

 

Keywords: Welfare, Expenditure, Percentage of Povery, Birespon Spline, Truncated, MSE

Keywords: Welfare, Expenditure, Percentage of Povery, Birespon Spline, Truncated, MSE

Article Metrics: