The ghost of unbalanced species distribution data in geographical model predictions

By A. Jiménez‐Valverde, Jorge M. Lobo. 2006

119 citations recorded by the source corpus.

Subfields: Ecological Modeling, Ecology, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/0879441c5d47038fd1ed62ac1009f4d7793fb96a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Ecological Modeling

In Ecology

In Nature and Landscape Conservation

Citations in the corpus

Not as good as they seem: the importance of concepts in species distribution modelling (2008, background): “…and discussing important conceptual and methodological aspects of species distribution models (e.g. Soberón & Peterson, 2005; Araújo & Guisan, 2006; Jiménez-Valverde & Lobo, 2006, 2007b; Kearney, 2006; Real et al ., 2006; Austin, 2007; Soberón, 2007; Lobo et al ., 2008; Raes & ter Steege, 2008).”
Prevalence, thresholds and the performance of presence–absence models (2014, background): “It is widely believed that biased predictions can be avoided by adopting different costs depending on the observed prevalence (Jim enez-Valverde & Lobo 2006; Lobo, Jim enez-Valverde & Real 2008), but an explicit consideration of costs shows why this is incorrect.”
More complex distribution models or more representative data (2008, methodology): “…the validation data, percentages of presences and absences correctly predicted (sensitivity and specificity scores) were also calculated after applying to model probabilities the threshold that minimizes the difference between sensitivity and specificity (see Jiménez-Valverde & Lobo 2006 and 2007).”
Threshold criteria for conversion of probability of species presence to either–or presence–absence (2007, methodology): “Hence, we recommend the MDT threshold criteria as the one of more general use (see also Liu et al., 2005; Jiménez-Valverde and Lobo, 2006).”
Modeling Species Distribution and Change Using Random Forest (2011, background): “However, Jiménez-Valverde and Lobo (2006) imply that unbalanced samples are not as serious a problem as historically thought.”
The uncertain nature of absences and their importance in species distribution modelling (2010, methodology): “The prevalence of the training data (0.1) was used as the threshold to convert these continuous maps into Boolean presence/absence maps (Jiménez-Valverde and Lobo 2006, 2007a).”
Equilibrium or not? Modelling potential distribution of invasive species in different stages of invasion (2012, methodology): “As the threshold selection and evaluation of predictive models is more influenced by prevalence than sample size (McPherson et al., 2004; Jimenez-Valverde & Lobo, 2006), a threshold of 0.5 prevalence was used, yielding 500 presence and 500 absence validation points sampled from the true…”
Exploring the effects of quantity and location of pseudo-absences and sampling biases on the performance of distribution models with limited point occurrence data (2011, methodology): “…to a true presence point, one must use an appropriate cut-off value to convert decimal fraction probabilities to a binary variable (Liu et al. 2005; Jiménez-Valverde & Lobo 2006; Jiménez-Valverde & Lobo 2007), or re-scale logistic probabilities by applying a favourability function (Real et al.…”
AUC: a misleading measure of the performance of predictive distribution models (2008, methodology): “It is necessary to rescale these probabilities if we want to use continuous and not binary predictions (Jiménez-Valverde & Lobo, 2006; Real et al .”
Applications of species distribution modeling to paleobiology (2011, background): “Suominen, 1972e1994, and triangles, digitized from Jonsell (1999)).”