Accuracy enhancement of last-mile delivery in Bali using fuzzy matching approaches for Banjar-based addresses

This research addresses the challenges of last mile delivery accuracy in Gianyar Regency, Bali, Indonesia by applying fuzzy matching techniques to Banjar-based addresses, which often suffer from variations and inconsistencies. These problems due to cultural and social factors. As an impact, around 32% of deliveries in a leading shipping company (e.g. Pos Indonesia) is failed due to variations and inaccuracies in the written address which usually Banjar-based. In Balinese culture, A Banjar in Bali is a customary community unit consisting of residents in a specific neighborhood or customary village, serving as the center for social, cultural, and religious activities, managing daily life including customs, ceremonies, and residential addressing. The research applies three fuzzy matching methods Levenshtein Distance, Token Sort Ratio, and Partial Ratio in order to match between the written address and standard address. Among these three approaches, Token Sort Ratio achieved the highest average similarity score (85.6), indicating superior performance in handling address variations. The study further analyzes delivery success across districts, identifying areas with relatively easy or difficult delivery conditions. By leveraging fuzzy matching, the research provides a data-driven approach to improve address matching accuracy, thereby enhancing delivery efficiency and reliability. These findings offer a better strategy for developing adaptive and efficient courier management and logistics business solutions to optimize last mile delivery in regions with generally non-standardized address writing.