STUDI PERBANDINGAN METODE STEPWISE DENGAN METODE SEMUA KEMUNGKINAN REGRESI UNTUK MENENTUKAN PERSAMAAN REGRESI TERBAIK

Abstract

Linier connection between X limited variable with Y fied variable can be formulased in fungsional relation form by regression analysis. There are some method to decide the best regression equation such as stepwise method, all probabilities method and Press method (Prediction Sum of Squares), etc. the purpose of this research is to compare stepwise method and all probabilities method in their efficiency aspect in order to decide the best regression equation. On stepwise method it done by put the variable one by one in the model until regression equation thet have real influence in explain the data is founded. While in the other hand, in all probabilities method it done by make all regression equation probabilities based on all unfied variable. The data that used in this research is searching data that obtainforum "Jawa Timur on 2001" book, published by BPS, active KB and contraception tools with unfied variable which are, X1(AKDR/IUD), X2(Tablet/pil), X3(Condom), X4(Male), X5(Female), X6(Injection), X7(Susuk), X8(Medicine) and fied variable Y (Couple's Fertile Period / pasangan usia subur). Based on analysis result with step wise method as the comparison is R2 and for all probabilities method as the comparison are R2 CP stat, both method have means and same result with equation :

Y = 21386,261 + 1,205 X1 + 2,142 X2 + 2,915 X4 + 3,327 X5 + 5,846 X6

Determination coefficient (R2) = 0,942 = 94,2%