Bayesian Networks for Agrotourism Success in Albania

Albania's agrotourism sector has grown rapidly, with over 270 certified establishments generating more than €12 million annually. Yet, understanding the determinants of success remains limited. This paper introduces a probabilistic framework using Bayesian Networks to model and predict agrotourism success. We define a directed acyclic graph connecting structural, spatial, operational, digital, and economic factors, with the joint distribution factorizing as \(∏ P(X_i Pa(X_i))\). Using data from 40+ agrotourism centers, we derive closed-form estimators, sensitivity measures, and intervention effects via do-calculus. Results indicate that rating, distance, and centrality are the most influential factors, and combined interventions can raise success probability by up to 71%. The framework offers evidence-based support for policymakers and operators.