Data-driven fleet optimization using ML algorithms and a decision-making grid framework

The highest impact on the cost center for fleet management is held by maintenance expenses and fuel consumption. The traditional ways of monitoring fleet performance fail to connect the raw operational data with the driving habits. The current research addresses this challenge by developing an architecture of frameworks, consisting of unsupervised and supervised machine learning algorithms, statistical testing, simula-tion and survival analysis to discover insights that lead to the key behavioral predic-tors. The nucleus of this complex architecture was the Decision-Making Grid (DMG), a two-dimensional matrix that groups vehicles based on their frequency of entering the service and the cost of the repairs. It is the first integration of DMG with ML for pre-scriptive fleet management. The objective of the study is twofold: firstly, to build a sys-tem that classifies the vehicle according to their risk profile and secondly to offer clear directions for changing the driver patterns that mostly affects the vehicle costs or keeping the good practices. The framework proposed by the research not only drives to optimization of operational efficiency but also contributes to a methodology that links driver profile to the cost, offering a scalable methodology for similar business contexts.