A multi-species approach to fisheries management requires taking into account the interactions between species in order to improve recruitment forecasting. Recent advances in Bayesian networks direct the learning of models with several interrelated variables to be forecasted simultaneously. These are known as multi-dimensional Bayesian network classifiers (MDBNs). Pre-processing steps are critical for the posterior learning of the model in these kinds of domains. Therefore, in this study, a set of 'state-of-the-art' uni-dimensional pre-processing methods, within the categories of missing data imputation, feature discretization and subset selection, are adapted to be used with MDBNs. A framework that includes the proposed multi-dimensional supervised pre-processing methods, coupled with a MDBN classifier, is tested for fish recruitment forecasting. The correctly forecasting of three fish species (anchovy, sardine and hake) simultaneously is doubled (from 17.3% to 29.5%) using the multi-dimensional approach in comparison to mono-species models. The probability assessments also show high improvement reducing the average error (Brier score) from 0.35 to 0.27. These differences are superior to the forecasting of species by pairs.
[1]
Momme Butenschön,et al.
Modelling the effects of climate change on the distribution and production of marine fishes: accounting for trophic interactions in a dynamic bioclimate envelope model
,
2013,
Global change biology.
[2]
José Antonio Lozano,et al.
A general framework for the statistical analysis of the sources of variance for classification error estimators
,
2013,
Pattern Recognit..
[3]
Gunnar Stefansson,et al.
The potential use of a Gadget model to predict stock responses to climate change in combination with Bayesian networks: the case of Bay of Biscay anchovy
,
2011
.
[4]
Jose A. Lozano,et al.
Fish recruitment prediction, using robust supervised classification methods
,
2010
.
[5]
Laura Uusitalo,et al.
Advantages and challenges of Bayesian networks in environmental modelling
,
2007
.
[6]
Linda C. van der Gaag,et al.
Multi-dimensional Bayesian Network Classifiers
,
2006,
Probabilistic Graphical Models.