Maritime transportation system, economy distribution and seaport network efficiency using fuzzy data envelopment analysis with clustering approach / Dineswary Nadarajan

Dineswary , Nadarajan (2023) Maritime transportation system, economy distribution and seaport network efficiency using fuzzy data envelopment analysis with clustering approach / Dineswary Nadarajan. PhD thesis, Universiti Malaya.

	PDF (The Candidate's Agreement) Restricted to Repository staff only Download (290Kb)
	PDF (Thesis PhD) Download (4Mb)

Abstract

The maritime transportation system is investigated by considering the marine traffic flow passing through Westport, Malaysia along the Strait of Malacca as a local case study. Firstly the Westport’s operational scheduling at the yard and land sides are explained and proposed. Next, dispersing and merging functions based on single junction theory that defined the marine traffic flows through the Westport are proposed so that the final macroscopic model describing that local maritime transportation system can be developed for the first time. This study utilizes multiple methods commonly used in assessing the maritime economy distribution where it is found that the import economy has more equality as compared to the export economy. Distance to Competitive Balance (DCB) has firstly applied in the thesis to determine the market concentration of 15 top leading import and export economies of the world. Tobit regression and data envelopment analysis (DEA) are conducted in seaport network efficiency measurement of 133 countries using LSCI as one of the output variable. In order to overcome the uncertainty in the real data, fuzzy DEA (FDEA) is performed by utilizing triangular fuzzy number (TrFN) and trapezoidal fuzzy number (TpFn) in the DEA calculation where the result comparisons have been done. As part of the present study’s original contribution, fuzzy linear regression modelling is also explored to highlight the interval-based regression technique using Possibilistic Linear Regression Least Squares (PLRLS) method. PLRLS determines the interval of minimum and maximum seaport network efficiency scores which gives better estimation overview of the score bounds than the regular regression model. Moreover, the unsupervised k-means, hierarchical and hierarchical k-means (hkmeans) strategies are imposed on the DEA and FDEA datasets of the seaport network efficiency scores. Clustering results between the three strategies are analysed and compared. Here, 133 global seaport countries are fitted into four efficiency clusters newly introduced in this thesis, namely low connectivity (LC), medium connectivity (MC), high connectivity (HC) and very high connectivity (VHC). Finally the hkmeans strategy is proposed as the best strategy for the seaport network efficiency clustering due to better countries composition in the four clusters and due to hkmeans strategy eliminates the drawback issues in the k-means and hierarchical clustering strategies.

Actions (For repository staff only : Login required)