Forecasting antimicrobial resistance (AMR) is critical for public health, yet most models neglect the interconnected nature of agricultural systems. Focusing on ciprofloxacin resistance in Campylobacter jejuni—a leading foodborne pathogen in poultry—this study aims to develop a probabilistic framework for identifying high-risk environmental conditions. We employed a graph-based machine learning and Bayesian approach, integrating and discretizing data from international databases. An exploratory classification with XGBoost and SVC was followed by core analysis using a Generalized Naive Bayes (GNB) model for feature selection and a Bayesian Network (BN) to uncover conditional dependencies. The GNB model identified pesticides, land use, and precipitation as key features. The BN revealed a complex web of interactions, showing that resistance probability is highly context-dependent. Precipitation was a critical effect modifier; for example, expanded land use correlated with an 18.3% increase in resistance probability during dry conditions but a 73.7% decrease during wet periods. Scenarios with low and high precipitation were associated with high risk, indicating multiple environmental pathways. Our results demonstrate that Bayesian networks can effectively model the complex, non-linear relationships driving AMR. Ciprofloxacin resistance emerges from system-wide interactions rather than isolated factors. This approach provides a valuable framework for generating hypotheses and supports the development of early-warning systems for targeted antimicrobial stewardship in poultry production.

Vine copula models have become highly popular and practical tools for modelling multivariate probability distributions due to their flexibility in modelling different kinds of dependences between the random variables involved. However, their flexibility comes with the drawback of a high-dimensional parameter space. To tackle this problem, truncated vine copulas were introduced by Kurowicka (2010) (Gaussian case) and Brechmann and Czado (2013) (general case). Truncated vine copulas contain conditionally independent pair copulas after the truncation level. So far, in the general case, truncated vine constructing algorithms started from the lowest tree in order to encode the largest dependences in the lower trees. The novelty of this paper starts from the observation that a truncated vine is determined by the first tree after the truncation level (see Kovács and Szántai (2017)). This paper introduces a new score for fitting truncated vines to given data, called the Weight of the truncated vine. Then we propose a completely new methodology for constructing truncated vines. We prove theorems which motivate this new approach. While earlier algorithms did not use conditional independences, we give algorithms for constructing and encoding truncated vines which do exploit them. Finally, we illustrate the algorithms on real datasets and compare the results with well-known methods included in R packages. Our method generally compare favorably to previously known methods.

Vine copulas are a flexible way for modeling dependences using only pair-copulas as building blocks. However if the number of variables grows the problem gets fast intractable. For dealing with this problem Brechmann at al. proposed the truncated R-vine copulas. The truncated R-vine copula has the very useful property that it can be constructed by using only pair-wise copulas, and conditional pair-wise copulas. In our earlier papers we introduced the concept of cherry-tree copulas. In this paper we characterize the relation between the cherry-tree copulas and the truncated R-vine copulas. Both are based on exploiting of some conditional independences between the variables. We give a necessary and sufficient condition for a cherry-tree copula to be a truncated R-vine copula. We introduce a new perspective for truncated R-vine modeling. The new idea is finding first a good fitting cherry-tree copula of order $k$. Then, if this is also a truncated R-vine copula we apply the Backward Algorithm introduced in this paper. This way the construction of a sequence of trees which leads to it becomes possible. So the cherry-tree copula can be expressed by pair-copulas and conditional pair-copulas. In the case when the fitted $k$ order cherry-tree copula is not a truncated R-vine copula we give an algorithm to transform it into truncated R-vine copula at level $k+1$. Therefore this cherry-tree copula can also be expressed by pair-copulas.

Vine copulas can efficiently model multivariate probability distributions. This paper focuses on a more thorough understanding of their structures, since in the literature, vine copula representations are often ambiguous. The graph representations include the original, cherry and chordal graph sequence structures, which we show equivalence between. Importantly we also show a new result, namely that when a perfect elimination ordering of a vine structure is given, then it can always be uniquely represented with a matrix. O. M. N\'apoles has shown a way to represent vines in a matrix, and we algorithmify this previous approach, while also showing a new method for constructing such a matrix, through cherry tree sequences. We also calculate the runtime of these algorithms. Lastly, we prove that these two matrix-building algorithms are equivalent if the same perfect elimination ordering is being used.