UPSI Digital Repository (UDRep)
|
UPSI Digital Repository (UDRep)
|
|
|
|
||||||||||||||||||||||||||
| Abstract : Perpustakaan Tuanku Bainun |
| This research aimed at developing Libyan energy generation models. Specifically, the research focused on the problems of outliers, non-normal and mullticollinearity. The presence of multivariate outliers was determined using Mahalanobis distance (MD) method. The non-normal of the data distribution was tested using a Mardia's test of multivariate skewness and kurtosis. While the multicollinearity of the data were determined using two methods, namely (i) correlation coefficient, and (ii) Variance Inflation Factor (VIF). The robust partial least squares-structural equation modelling (RPLS-SEM) was utilized to develop Libyan energy generation model.The research proposed the use of an alternative robust measure of central tendency, namely the Adaptive Winsorized Mean (AWM), which used winsorization for determining the data distribution as symmetrical or asymmetrical. The results from RPLS-SEM analysis revealed that each block in the developed model consisted of strong latent variables. The results also showed that the standard measures of reliability, validity and model fit represented their respective latent constructs adequately. In addition, an overall evaluation of the structural model through Goodness of Fit (GoF) index indicated that the model fits the data well. The performance comparison between conventional Partial Least Squares-Structural Equation Model (PLS-SEM) and RPLS-SEM showed that the RPLS-SEM was more effective than the PLS-SEM. In conclusion, the research has successfully developed and evaluated the Libyan energy generation model using RPLS-SEM in the presence of outliers, non-normal and multicollinearity of data. In implication, the new proposed robust model can sustain effectively and also helps to achieve robust predictions. |
| This material may be protected under Copyright Act which governs the making of photocopies or reproductions of copyrighted materials. You may use the digitized material for private study, scholarship, or research. |