A Data-Driven Decision-Making Framework for Fleet Management in the Government Sector of Dubai
Loading...
Date
2024-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
The British University in Dubai (BUiD)
Abstract
Data-driven decision-making has become increasingly widespread and relevant across all business areas, including private and public sectors. My research aims to develop a data-driven decision support framework for fleet management, focusing on leveraging advanced algorithms, including decision trees and random forests, to generate domain-specific AI models. These models are applied on top of ISO 55001 to create A data-driven decision-making framework for fleet management in the government sector of Dubai. The proposed framework comprises key elements: Important Decisions derived from interviews with transportation leaders, Knowledge Management enhanced by AI algorithms, Data Mining/Analysis utilizing historical data, the Fleet Management System employing Oracle ERP, and a Data-Driven Decision Support Framework that leans towards the extended framework approach. The framework is validated through three domain expert interviews; the insights gleaned from the experts have instilled a sense of optimism regarding its theoretical efficacy and its capacity to adeptly tackle the distinct challenges encountered within the realm of government fleet management in Dubai. The approach involved 12 interviews with Dubai Municipality's fleet managers, an analysis of industry standards, and a literature review. Key decisions, including predicting vehicle cancellations and developing a maintenance plan, were identified. A prediction model for vehicle cancellation, using the Random Forest Classifier, demonstrated high accuracy (F1 score of 0.8824). Additionally, an AI model predicting heavy vehicle failures with gradient-boosted trees achieved 68.09% accuracy.
Description
Keywords
data-driven decision-support, framework, AI, fleet management