PAPER TITLE :HETEROSCEDASTIC D-OPTIMAL DESIGN IN REGRESSION MODELS WITH APPLICATION IN KINEMATIC VISCOSITY DATA

FUTA JOURNAL OF RESEARCH IN SCIENCE | VOLUME 15 NUMBER 2 2019

Paper Details

  • Author(s) : Bodunwa O.K*, Fasoranbaku O.A and Ajiboye. A. S.
  • Abstract:

In real life situations, the assumption of homogeneity is often violated and the variances of the error terms are not the same which is termed heteroscedasticity. In this work, D-optimality criterion was used when there is heterocedsasticity in the data set and when the data had been corrected using different methods for correction thereby making the variance of the error structure to be equal. Comparing the result for the two models used, when there is heteroscedasticity and when it has been corrected, the variances and the determinants of dispersion matrices shows that D-optimal design when the data set has been corrected is more efficient than when there is heteroscedastic

Key words: D-optimal, Heteroscedasticity, Experimental Design, Regression model