APLIKASI ANALISIS FAKTOR DENGAN METODE PRINCIPAL COMPONENT ANALYSIS DAN MAXIMUM LIKELIHOOD DALAM FAKTOR-FAKTOR YANG MEMENGARUHI PEMBERIAN MAKANAN TAMBAHAN PADA BAYI USIA 0-6 BULAN DI DESA PEMATANG PANJANG KECAMATAN AIR PUTIH KABUPATEN BATUBARA TAHU

iska simarmata, Abdul Jalil Armi Arma, arnita .

Abstract


Factor analysis is one of the multivariate statistical analysis techniques.This analysis is included in the interdependence technique with the aim of reconciling data in a grouping or the formation of a new set of variables which is named factor. The parameter estimation that is commonly used in this analysis is the principal component analysis method and the maximum likelihood method.

This research aims to know the comparison of suitability of the model by principal component method and maximum likelihood method within the factors that affect the complementary feeding in infants ages 0-6 months in Pematang Panjang Village Air Putih Subdistrict Batubara District 2013. Based on its purpose, this research is implementative research and based on its explanation level it is comparative research. The population of the research was all mothers who have baby in age of 0-6 months which are as many as 52 persons.  The sampleis all population made as sample.

The result of factor analysis using the principal component analysis method forms factor 1 (education, culture, economy, job, and mother’s health) and factor 2 (knowledge, baby’s health, and health/medical officer), while the result of factor analysis using maximum likelihood method forms factor 1 (education, culture, economy and job) and factor 2 (knowledge, baby’s health, mother’s health and health/medical officer).

Research results by using analysis of factors suggest that the maximum likelihood method has a better model accuracythan the principal component analysis method, because the RMSE value of maximum likelihood method which is 0,0222 < RMSE value of principal component analysis method which is 0,0409.

It is suggested to the next research which uses factor analysis aplication that it is better to firstly see the result of the analysis using principal component analysis and maximum likelihood methods and then using method with less RMSE value.

Full Text: PDF

Refbacks

  • There are currently no refbacks.