TY - JOUR
TI - PENENTUAN KOEFISIEN KORELASI KANONIK DAN INTERPRETASI FUNGSI KANONIK MULTIVARIAT
AU - Asbah, Muhamad Faliqul; Sudarno, Sudarno; Safitri, Diah
IS - Vol 2, No 2 (2013): Wisuda Periode April 2013
PB - Jurusan Statistika UNDIP
JO - Jurnal Gaussian
PY - 2013
SP - 119
EP - 128
UR - http://ejournal-s1.undip.ac.id/index.php/gaussian/article/view/2778
AB - Canonical correlation analysis is a useful technique to identify and quantify the linier relationships, involving multiple independent and multiple dependent variable. It focuses on the correlation between a linier combination of the variables in one set independent and a linier combination of the variables in another set dependent. The pairs of linier combinations are called canonical function, and their correlation are called canonical correlation coefficient. The statistical assumptions should be fulfilled are: linearity, multivariate normality, homoscedasticity, and nonmulticollinearity. The use of variable consists of three dependent variable: y1 =Maximum daily relative humidity, y2 = Minimum daily relative humidity, and y3 = Integrated area under daily humidity curve and three independent variable: x1 = Maximum daily air temperature, x2 = Minimum daily air temperature, and x3 = Integrated area under daily air temperature curve. For The result of canonical correlation analysis indicate that there are two significant canonical correlation between the daily air temperature level with the daily humidity level. The reduncancy index showed that the daily humidity level can explained a total of 69 % of the variance in the daily air temperature level, otherwise the daily air temperature level can explained a total 60 % of the variance in the daily humidity level. Interpretations involves examining the canonical function to determine the relative contibution of each of the original variables in the canonical relationships: canonical weights, canonical loadings, and canonical cross loadings showed that the sequence variables which contribute on the independent variate are x1,x3, and x2. Then, the sequence variables which contribute on the dependent variate are y1, y2, and y3.