Cosic, Jelena
Bayesian Network as a Modelling Tool for Increasing Knowledge on the Factors Influencing Vineyard Longevity and Sustainability
Jelena Cosic1, Steffen Klaere1, Matthew Goddard2 and Bruno Fedrizzi3
1. Department of Statistics and School of Biological Sciences, University of Auckland
2. School of Biological Sciences, University of Auckland
3. School of Chemical Sciences, University of Auckland
The long-term project “Resilient and Profitable NZ wine industry” has the objective to study the impact of different vineyard management techniques on the vineyard longevity and profitability, and to increase the knowledge of the factors influencing longevity and profitability. To find meaningful answers appropriate quantifiable outcomes need to be obtained. Profitability of a vineyard can be quantified by its yield and quality of the end product, while health will be studied in a more holistic way by developing a vineyard ecosystems model incorporating the data obtained from different areas of interest. The empirical nature of data collection makes a computational ecosystem modelling approach the most suitable. Such approaches are quite common and popular in ecology, and are promising for this project. Of particular interest are Bayesian Networks (BNs) which have received increased attention throughout several research fields for their ability to incorporate prior knowledge and to handle incomplete data. BN have also been shown to efficiently avoid overfitting the data, and avoiding the observation of “chimeric” effects. We will use BN to model vineyard ecosystems incorporating microbial, fungal and eukaryotic molecular data, chemical profiles, meteorological information, and other markers at different points in the life cycle of vineyards, and discover the differences vineyard managements make with respect to resilience and profit. Some of the challenges that we see are: variables that have been measured on different time scales, a large amount of microbial data and uncertainty of the interactions of components included in our ecosystem.
This presentation is eligible for the NZSA Student Prize.