[ABE-L] INSPER DATA SCIENCE ACADEMIC SEMINAR

[Hedibert Lopes]

*A Bayesian Network approach to modelling food security over time*

*Speaker:* Thais Fonseca <https://sites.google.com/site/thaisf/>

*University:* Universidade Federal do Rio de Janeiro - UFRJ
<https://ufrj.br/> <https://web.mit.edu/>



Chronic food insecurity represents one of the main obstacles to economical
and social development in many countries. Very often, Policymakers struggle
to improve the food security scenario due to the low budget for
investment.  In ever-larger dynamic systems, such as the food system, it is
increasingly difficult for decision-makers to effectively account for all
the variables within the system that may influence the outcomes of interest
under enactments of various candidate policies. For instance, the food
system is usually influenced by food prices, availability, income,
unemployment and demographic factors to cite a few. Each of the influencing
variables is likely to be dynamic sub-systems with expert domains supported
by sophisticated probabilistic models. This work integrates historical data and
expert knowledge within the framework of graphical models to allow for
modelling and predicting food security over time. The proposed tool aims to
provide a mechanism to highlight the most relevant factors for the
improvement of the system, allowing Policymakers to make the optimal
decision in the presence of uncertainty and cost restrictions. The model is
based on Dynamic Bayesian Networks which are computational efficient
providing fast risk evaluation for several competing policies. An
application to food security in the UK is presented at the household level
which combines both economical, meteorological and epidemiological data and
expert information in a comprehensive model for malnutrition and school
attainment. Finally, an extension to model food security in Brazil using
the PNAD data and dynamical Bayesian Networks for multivariate counts is
presented.

September 23, 2021, 12pm-1:20PM de São Paulo, Brasil
Zoom link: https://zoom.us/j/95660781314
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <http://lists.ime.usp.br/pipermail/abe/attachments/20210922/b9d24830/attachment.htm>