Open AccessIdentification of correlation structure using rotated factor loadings
Author(s) -
I.A. Iwok,
Nwikpe B. J
Publication year - 2017
Publication title -
international journal of advanced statistics and probability
Language(s) - English
Resource type - Journals
ISSN - 2307-9045
DOI - 10.14419/ijasp.v5i1.6931
Subject(s) - factor analysis , statistics , mathematics , principal component analysis , unobservable , factor (programming language) , correlation , medicine , econometrics , computer science , geometry , programming language
This work seeks to identify the correlation structure of variables in terms of few underlying but unobservable factors. The method was applied to age and five different tests results obtained from 200 patients in a hospital. Two factors were identified using the scree plot and the Kaiser criterion. The factor loadings obtained by the method of principal components gave an inadequate fit to the data. An algebraic approach was applied using orthogonal rotation, and the loadings were found to give a clear and interpretable pattern. Consequently, the variables: age, fasting blood sugar and diastolic blood pressure were found to cluster about the first factor F1 called Age-Cardiovascular factor. Similarly, the remaining variables malaria, typhoid and haemoglobin clustered about the second factor F2 and the given name was Hemo-typhomalaria factor. Diagnostic checks were carried out and the factor model generated by the rotated loadings was found to be adequate.